Introduction

The Hyrcanian floral province stretches along the south of the Caspian Sea, ranging from northern Iran to southeastern Azerbaijan (Zohary 1973). Mean annual temperatures of about 17°C and an annual rainfall of 1,000–1,500 mm have given rise to dense deciduous forests that extend from the Caspian lowlands to about 2,500 m altitude on the slopes of the Elburs Mountains. The region exhibits an exceptional wealth of wild plant species and has been considered as an evolutionary centre for fruit trees (Khoshbakht and Hammer 2005). However, recent urbanization and human disturbance have caused increasing deforestation and degradation, with a concomitant threat to biodiversity (Scharnweber et al. 2007).

Cherry plum, Prunus divaricata Ledeb., is a wild growing, diploid, self-incompatible fruit tree that belongs to section Prunus, subgenus Prunus, within the family of Rosaceae (Faust and Surányi 1999; Scholz and Scholz 1995; Reales et al. 2010). Pollination is entomophilous, and seeds are dispersed by frugivorous animals. The species is widely distributed from the Balkan Peninsula across Anatolia and the Caucasus to Central Asia, including the Hyrcanian forests of northern Iran (Browicz 1969, 1997). Individual trees have been semi-cultivated for their edible fruits, especially in home gardens along the Caspian coast (Khoshbakht et al. 2007). The fresh fruits with ~2–4 cm in diameter are part of the diet of the local people, and are either eaten raw or used to prepare a local tart candy called “Lavashak”. The species is also widely used as a rootstock (Scholz and Scholz 1995) and certainly has some potential for further domestication, providing economic and livelihood benefits for subsistence farmers.

One can find two alternative names for this species, i.e., Prunus divaricata Ledeb. and Prunus cerasifera Ehrh. (Browicz 1997; Büttner 2001). The latter name is often used for cultivated forms, which are also referred to as myrobalan plums. According to Browicz (1997) wild forms (Prunus cerasifera ssp. divaricata) and cultivated forms (Prunus cerasifera ssp. cerasifera) should however, rather be distinguished at the level of subspecies. A considerable array of additional subspecies and varieties have been recognized from various regions of the natural distribution range of cherry plum, but numerous transitional characters render the distinction of these entities difficult (Browicz 1969; Büttner 2001). Eremin and Garkovenko (1989) proposed to separate the species into three subspecies, namely P. cerasifera ssp. cerasifera (syn. Prunus divaricata Ledeb.), P. cerasifera ssp. orientalis (Koval.) Erem. et Garkov and P. cerasifera ssp. macrocarpa (Erem. et Garkov), with all the cultivated forms included in the latter. This proposal was adopted by Büttner (2001) but is nevertheless provisional and needs further taxonomic evaluation in the light of molecular data.

Whereas the demand for cherry plums is growing steadily, their supply from the wild is threatened by deforestation. To conserve and protect such valuable plant material, optimized breeding and domestication programs are compulsory. During the last years, the cultivated form of cherry plum was in the focus of genetic investigations because several of its clones (e.g., P.2175 and P.2980) are highly resistant to root-knot nematodes of the genus Meloidogyne (Dirlewanger et al. 2004; Lecouls et al. 2004). The cultivated myrobalan was also suggested as a useful diploid model system for studying the molecular genetic background of self-incompatibility in plums (Sutherland et al. 2009). However, no attempts have been made so far to evaluate population genetic parameters of its wild relative P. divaricata and to assess the possible effects of destruction, fragmentation, and conversion of natural habitats on its genetic variability. These questions are of particular interest in a fast-developing region like northern Iran, where local people take nutritional advantage of the species.

In the last two decades, genetic information has contributed substantially to both cultivation and conservation biology of plants and animals on a worldwide scale. Molecular fingerprints based on anonymous, PCR-based molecular markers such as AFLP (amplified fragment length polymorphism, Vos et al. 1995) have proven to be efficient tools for detecting genetic relationships and genetic diversity, and were also used to elucidate population structure, gene flow and rare genotypes in plants (e.g., Panaud et al. 2002; Peters et al. 2009). However, these anonymous markers are normally inherited in a dominant fashion, which limits their applicability for population genetics. Nuclear microsatellites, also called simple sequence repeats (SSRs), are currently the genetic markers of choice in population studies and for the assessment of genetic diversity and differentiation (Balloux and Lugon-Moulin 2002; Powell et al. 1996). Microsatellites consist of tandemly repeated, short DNA sequence motifs and are frequently size-polymorphic in a population due to a variable number of tandem repeats. Moreover, they are ubiquitous components of eukaryotic genomes and can be found both in coding and non-coding regions. The popularity of nuclear microsatellites stems from a unique combination of several important advantages, namely their Mendelian and codominant inheritance, high abundance, enormous extent of allelic diversity, and the ease of assessing size variation by PCR with pairs of flanking primers. The only serious disadvantage is the necessity of sequence information for primer design. Introduction of library enrichment techniques and automatic sequencing have simplified their expensive isolation, thus allowing wide application in plant genetics (Weising et al. 2005). Moreover, flanking regions of SSRs are often conserved in related species, which enables the use of the same primer pairs in related genomes (“cross-species amplification”).

An increasingly important source for microsatellite markers are expressed sequence tag (EST) databases (Pashley et al. 2006; Ellis and Burke 2007). EST-derived SSRs combine several important advantages. First, in silico mining for EST-SSRs is fast and easy, compared to standard cloning and sequencing procedures. Second, ESTs-SSRs are physically linked to a gene, and putative gene functions can readily be identified by a BLAST comparison with protein databases (Yao et al. 2010). Third, primer target sequences that reside in transcribed regions are expected to be relatively conserved, thus enhancing the chance of marker transferability across taxa (Decrooq et al. 2003; Gasic et al. 2009; Vendramin et al. 2007). On the negative side, the association with coding regions may limit the polymorphism of EST-derived microsatellite markers (Ellis and Burke 2007).

The present study aims to investigate the population genetics of wild Prunus divaricata using nuclear SSRs. Whereas large numbers of SSR markers are already available from other Prunus species such as peach (P. persica; Aranzana et al. 2002; Cipriani et al. 1999; Dirlewanger et al. 2002), almond (P. dulcis; Mnejja et al. 2005; Xie et al. 2006), apricot (P. armeniaca; Messina et al. 2004) and Japanese plum (P. salicina; Mnejja et al. 2004), none have been developed yet for P. divaricata. In the first part of our study, we therefore tested the transferability of 18 anonymous and 7 EST-derived SSR markers originally developed for peach and Japanese plum to P. divaricata. The eleven best-performing heterologous markers were then used to assess the genetic variability and differentiation of wild P. divaricata within and among 12 geographically separated populations along the Iranian coast of the Caspian Sea.

Materials and methods

Plant material and DNA extraction

Leaves were collected from 117 wild P. divaricata individuals from 12 populations in an area between the Caspian Sea and the Elburs mountain range within the provinces of Gilan and Mazandaran (Fig. 1). The population samples were collected from natural forests with a minimum of anthropogenic influence, with only four of these populations being located close to a larger city (Table 1). Elevations ranged from −19 m up to 970 m a.s.l. Five to 15 individuals were sampled per population. The geographical distances between pairs of populations ranged from 17 km up to 430 km. Immediately after collection, leaves were quick-dried in silica gel, transported to the laboratory and stored at −80°C until use. Total genomic DNA was extracted using a variant of the CTAB method (Weising et al. 1995). Populations were named according to cities close to the collection site. Two samples each of sweet cherry (Prunus avium L.; diploid) and blackthorn (P. spinosa L.; tetraploid) and one sample of domestic plum (P. domestica L.; hexaploid) were used as additional reference material for the primer transferability tests.

Fig. 1
figure 1

 Locations of P. divaricata populations. P1: Kashafi, P2: Astara, P3: Hashtpar, P4: Kiasar, P5: Asalem, P6: Fowman, P7: Lahijan, P8: Tonkabon, P9: Zirab, P10: Safiedkouh, P11: Amol, P12: Babol

Full size image
Table 1 Sampling localities and sample sizes of P. divaricata populations studied
Full size table

Microsatellite analyses

Initially, 25 microsatellite-flanking primer pairs originally developed for P. persica (L.) Batsch and P. salicina Lindl. (see Table 2) were tested for successful PCR amplification in seven randomly chosen individuals of P. divaricata and five samples of the reference material (see above). Fourteen primer pairs were derived from two genomic libraries of peach, enriched for AC/GT and AG/CT repeats (Cipriani et al. 1999; Testolin et al. 2000). Seven primer pairs were derived from peach EST-SSRs (Vendramin et al. 2007), and four were derived from a genomic SSR library of Japanese plum DNA enriched for CT repeats (Mnejja et al. 2004). Polymerase chain reactions (PCR) were performed in a total volume of 25 μl containing 1x PCR Mango-buffer (provided by the manufacturer, Bioline), 5 μg BSA, 1.5 mM MgCl2, 0.1 mM of each dNTP, 10 pmol of each primer, 10 ng of genomic DNA and 0.1 U of Taq DNA polymerase (Mango-Taq, Bioline). PCR cycling conditions consisted of an initial denaturation step of 94°C for 5 min, followed by 27 cycles of 45 s at 94°C, 45 s at 57°C and 45 s at 72°C, and a final extension step (8 min at 72°C). PCR products were separated by electrophoresis on 1.5% agarose gels (NEOO Ultra Quality, Roth) in 0.5x TBE (Sambrook and Russell 2001) at 10 V/cm, stained with ethidium bromide (1 μg/ml) and visualized under UV-light. A 100 bp DNA ladder (Roth) was used as molecular size standard. Candidate markers that passed these initial tests were used for analyzing the full sample set. The same cycling conditions were used, but one primer of each pair was 5′-labeled with IRD700 or IRD800 fluorescent dyes. Fluorescently labelled microsatellite fragments were analysed on a LiCor® IR2 DNA Sequencer Long Readir 4200 in high resolution polyacrylamide gels (6%), and allele sizes were determined by visual comparison with a size standard and a reference sample included in all gels.

Table 2 Results of cross-transferability tests of microsatellite markers from peach (UDP, EPPISF) and Japanese plum (CPSCT) amplified in cherry plum, sweet cherry, blackthorn and domestic plum
Full size table

Statistical analyses

Departures from Hardy–Weinberg equilibrium (HWE) at each locus and linkage disequilibrium between individual microsatellite loci were evaluated by an exact test using a Markov chain method and Bonferroni corrections implemented in GENEPOP (Raymond and Rousset 1995). Indications for the presence of null alleles were assessed using an iterative algorithm based on the observed and expected frequencies of the various genotypes by CERVUS (version 3.0.3, Kalinowski et al. 2007). To examine the informational content of each microsatellite locus the following parameters were calculated: allele frequency, number of alleles per locus, expected (Hexp, Nei 1987) and observed (Hobs, direct count estimate) heterozygosity for each population and each locus, Wright’s fixation indices (FIS, FIT, FST, Wright 1931, 1969, extended by Nei 1977) and Nei’s coefficient of genetic differentiation (GST, Nei 1973). All of these parameters were estimated with the software programs FSTAT (version 2.9.3.2, Goudet 2002), GENEPOP and/or GENETIX (version 4.05, Belkhir et al. 2000). The total genetic variation was partitioned into a between- and a within-population component by an analysis of molecular variance (AMOVA, Excoffier et al. 1992), using ARLEQUIN (version 3.11, Excoffier et al. 2005) with 1,000 permutations for significance testing. Genetic relationships among individuals were assessed by a multivariate principal component analysis (PCA), performed with GENETIX software using Euclidean distances between samples. The correlation between genetic distance and the corresponding geographic distances were analyzed using the Mantel test with 1,000 permutations (Mantel 1967) based on a pairwise matrix of Wright’s FST (also generated in ARLEQUIN).

Results

Cross-species transferability of microsatellite markers

The success of cross-species amplification of the 25 candidate markers in P. divaricata, P. avium, P. domestica and P. spinosa was evaluated by the quality of the banding patterns on agarose gels (Table 2). PCR amplification was considered successful when the number of distinct bands was compatible with the known ploidy status, and when bands of the expected size range were present in all Prunus samples. These criteria were met by 12 candidates. Of the remaining 13 loci, seven gave complex banding patterns, and five markers failed with one or more samples of the test set. No PCR product in any sample was obtained from locus UDP96-015. Eleven of the 12 successful candidates produced well-scorable, polymorphic bands also on PAA gels, whereas UDP96-003 yielded a monomorphic banding pattern on PAA gels and was excluded from further study.

Allelic variation

Eight genomic and three EST-derived microsatellite markers were selected for the population genetic analyses (Table 2). They detected 184 alleles (3–31 depending on the locus) in 117 individuals of P. divaricata sampled along the southern coast of the Caspian Sea, with a mean value of 16.7 alleles per locus (Table 3). No significant deviations from HWE were detected for any of the loci or populations, and no indications for linkage disequilibrium were found. Overall numbers of alleles were very similar across populations, varying between 5.1 and 8.5 with an average of 7.0. Rare alleles with frequencies <0.05 represented between 55 and 84% of the total allele spectrum, depending on the locus. Indications for the presence of null alleles were found in three markers (UDP96-008, UDP98-409, EPPISF018), but their estimated frequencies were very low (<0.05).

Table 3 Population genetic parameters determined for each of 11 microsatellite loci averaged across 12 populations of P. divaricata
Full size table

Genetic diversity

Levels of genetic diversity were generally high, with Hobs = 0.664 and Hexp = 0.733 averaged over all loci (Table 3) and Hobs = 0.682 and Hexp = 0.729 averaged over all populations (Table 4). In contrast, FST and GST values were low, ranging from zero (CPSCT012) to 0.102 (EPPISF018) with a mean of 0.029 over all loci for FST (P < 0.05) and from 0.047 (EPPISF016) to 0.139 (EPPISF018) with a mean of 0.079 over all loci for GST. The likewise low values of FIS and FIT calculated for each SSR locus in each population indicate that inbreeding is negligible, which is in line with the non-significant deviations from the HWE (see above). An analysis of molecular variance (AMOVA) indicated that only 3.2% of the total molecular variance was attributable to the divergence among populations. In contrast, 96.8% of the variance was found within populations (Table 5). Altogether, these results indicate a low level of genetic differentiation, and hence suggest high levels of gene flow between the studied populations of P. divaricata. The almost complete lack of differentiation was also reflected by a principal components analysis (PCA), where the first two axes together accounted for only 6.06% of the total genetic variability (pc 1 = 3.22%, pc 2 = 2.84%). The two-dimensional PCA diagram did not arrange the individuals into distinct geographical populations, as all 12 populations strongly overlapped (data not shown). A Mantel test nevertheless revealed a weak but significant correlation between geographic and genetic distances (r = 0.29, P < 0.003, Fig. 2).

Table 4 Observed (Hobs) and expected (Hexp) heterozygosities within each of 12 populations of P. divaricata averaged over 11 microsatellite loci
Full size table
Table 5 Results of the analysis of molecular variance (AMOVA) for 117 P. divaricata individuals grouped in 12 populations
Full size table
Fig. 2
figure 2

 Correlation of geographic distance (in kilometers) and genetic distance (pairwise FST) among 117 individuals of 12 populations of P. divaricata, including regression line (Mantel test, r = 0.29, P < 0.003)

Full size image

Discussion

Cross-species transferability of microsatellite markers within Prunus

Whereas no microsatellite markers have been developed so far in P. divaricata, numerous studies reported successful cross-species transferability of SSR markers among different Prunus species (e.g., Cipriani et al. 1999; Dirlewanger et al. 2002; Sánchez-Pérez et al. 2006; Wünsch 2009; Mnejja et al. 2010). For example, two markers from peach, BPPCT-007 (Dirlewanger et al. 2002) and CPPCT-006 (Aranzana et al. 2002) amplified in ten different Prunus species from three subgenera and five sections (Wünsch 2009). Marker transportability to other genera of the Rosaceae, like apple, pear and strawberry, seems to be much less efficient (Mnejja et al. 2010).

Based upon the close taxonomic relationship of P. divaricata with P. salicina (both from section Prunus) we expected relatively high levels of transferability for the CPSCT primers developed in the latter (Mnejja et al. 2004). Table 2 shows that indeed four out of four (100%) CPSCT primers amplified distinct and polymorphic PCR products in cherry plum, with 3–31 alleles. Much lower success rates were obtained with the UDP primers derived from the more distantly related peach (Cipriani et al. 1999), of which only 36% generated distinct PCR products in cherry plum. One marker, UDP96-003 only yielded a monomorphic band pattern on PAA gels. Of the EPPISF primers derived from peach ESTs (Vendramin et al. 2007), 71% gave a distinct and highly polymorphic PCR product in cherry plum, supporting the assumption that primers derived from transcribed sequences have an increased chance of cross-species functionality. With 31 alleles among 184 individuals, the EPPISF016 locus based on a trinucleotide repeat (CTT) motif in peach even turned out as one of the most polymorphic markers in cherry plum. This was quite unexpected given that trinucleotide repeats prevail in coding regions and are therefore supposed to be more conserved and therefore less polymorphic.

Genetic diversity

The 11 polymorphic microsatellite loci employed in the present study proved to be highly informative in P. divaricata and detected an average of 16.7 alleles per locus. This is a relatively large number compared to the average numbers of alleles per locus found in other studies dealing with different Prunus species, as e.g., 13.3 in P. armeniaca L. (Maghuly et al. 2005), 10.7 in P. cerasus L. (Cantini et al. 2001), 6.7 in P. mahaleb L. (Godoy and Jordano 2001), 8.2 in P. cerasoides D. Don (Pakkad et al. 2003), 7.3 in P. persica (L.) Batsch (Aranzana et al. 2003), and only 2.93 in Amygdalus nana L. syn. Prunus tenella Rehd. and A. (Tahan et al. 2009). The large number of alleles indicates a generally high level of genetic diversity within wild P. divaricata, which is supported by high values of observed and expected heterozygosity. For example, Hobs averaged over all loci and populations of. P. divaricata was 0.664, which is higher than 0.53 reported for wild P. cerasoides (Pakkad et al. 2003), a value that was considered by the authors as a high level of genetic variation. Mean values of Hobs obtained from cultivated Prunus species like apricot and peach ranged from 0.22 to 0.45 (Martínez-Gómez et al. 2003; Sánchez-Pérez et al. 2006), demonstrating that domesticated and cultivated species often show low levels of genetic variation as compared with their wild ancestors (Miller and Schaal 2006; Anthony et al. 2002; Cahill 2004; Panda et al. 2003).

Although the studied populations of P. divaricata represent only a small part of the total distribution range of this species in western Asia, the geographic distances between populations were relatively large. Nevertheless, the results of all analyses indicated that genetic differentiation between populations was very low. Both, AMOVA and PCA indicated that P. divaricata preserved the vast majority of its genetic variability within populations. A comparable but allozyme-based study of genetic variation in six wild P. avium L. populations reported by Mariette et al. (1997) likewise showed no recognizable structure and a mean GST value of 0.05, similar to our result (0.079). The authors concluded that these populations underlie both well-balanced selection and neutral genetic drift, respectively migration. Intensive research with allozyme markers has shown that seed dispersal patterns are among the main factors that determine the partitioning of genetic variation within and among populations (Hamrick et al. 1993; Hamrick and Godt 1997). The sweet and fleshy fruits of most fruit trees are dispersed by birds and mammals, leading to characteristically high levels of within-population genetic variation and low levels of among-population variation, due to large frugivorous birds acting as long-distance seed dispersers (e.g., woodpeckers, thrushes and pigeons). Long-distance flights away from the feeding trees have frequently been observed in these birds, e.g., in Prunus mahaleb (Jordano and Godoy 2000; Jordano and Schupp 2000). Likewise, anthropogenic influence may play a role for gene flow as the fruits of wild P. divaricata are part of the diet of local people in the fruiting season. Still another reason for the reduced genetic differentiation of P. divaricata may be associated with the geographical conditions of the area itself. The high elevations of the Elburs mountain range in the south and the Caspian Sea in the north may enhance the exchange of genetic material in east-west-direction.

Conservation aspects

The cultivated form of cherry plum, P. cerasifera Ehrh., is an agriculturally important species, and is distinguished from other plums through e.g. its drought tolerance, high resistance to root-knot nematodes and its suitability as a rootstock (Lecouls et al. 2004; Dirlewanger et al. 2004). Interestingly, however, the germplasm variability of its wild ancestor, P. divaricata, has neither been properly assessed nor collected, and essentially nothing was known so far about the genetic variability of this Prunus species. The results of our study indicate high levels of genetic diversity and a lack of genetic differentiation of P. divaricata in northern Iran, suggesting that natural populations of cherry plum in the Caspian forests are close to Hardy–Weinberg equilibrium and have not yet experienced measurable genetic drift. However, it cannot be excluded that deforestation, habitat fragmentation and the increasing population density in the area will have a negative impact onto the population structure of the species in the near future.

At any rate, it seems reasonable to secure the genetic variability present across the gene pool of wild cherry plum both in its natural habitats in northern Iran as well as ex situ in germplasm collections. A focus for in situ conservation should be placed on populations with the highest level of genetic diversity, such as the populations from Kashafi (P1), Hashtpar (P3), Asalem (P5) and Babol (P12). For ex situ conservation, seeds of P. divaricata should be collected from as many populations as possible and stored in a gene bank (Li et al. 2009). Wild species can be very useful in breeding programs as sources of genetic variability that enlarge the gene pool of their cultivated relatives (Wolko et al. 2010), and the highly diverse P. divaricata could well turn out to be a valuable gene donor to increase variation in cultivated Prunus species by cross-breeding (Fritz et al. 1994; Mandegaran et al. 1999; Dirlewanger et al. 2004).