Evidence of Low Chloroplast Genetic Diversity in Two Carpinus Species in the Northern Balkans

Genetic diversity and differentiation in two Carpinus species (C. betulus and C. orientalis) occurring in Romania was investigated by using three chloroplast Simple Sequence Repeat markers (cpSSRs). A total of 96 and 32 individuals were sampled in eighteen C. betulus and six C. orientalis populations, respectively. A total of four chloroplast haplotypes were observed. Two haplotypes were specific for C. betulus and two for C. orientalis. Most of C. betulus populations were fixed for the predominant haplotype (H1), which was observed in 82% of the individuals. All C. orientalis populations were fixed for one haplotype or the other. Populations with haplotype (H3) are spread in southern Romania and the haplotype (H4) was observed at the northern limit of C. orientalis natural distribution range. Genetic differentiation among populations was moderate in C. betulus (GST = 0.422), compared to the high value observed in C. orientalis (GST = 1.000), which can be explained by the occurrence of a distinct haplotype in the peripheral population. RST values for both species suggest low levels of recurrent gene flow through seeds among populations. Our data on geographic distribution of chloroplast DNA haplotypes may be useful for the identification and conservation of distinct genetic resources of the two Carpinus species.


Introduction
Genetic diversity is essential for species to evolve and to adapt to changing environmental conditions.Patterns of genetic diversity are influenced by population contractions and expansions during glacial-interglacial fluctuations and by post-glacial migration history (Petit et al., 2005).Phylogeographic studies have provided important evidence on species demographic history and population structure (Petit et al., 2005;White et al., 2007;Xu et al., 2015) revealing that species-specific life history traits and geographic barriers may have a strong influence on the current patterns of population genetic structure (Ortiz-Ramírez et al., 2016).
The genus Carpinus (Betulaceae family) is represented by two species in Europe (C.betulus L. and C. orientalis Mill).European hornbeam (C.betulus L.) is a temperate hardwood species especially in forests dominated by oak species, where it demonstrates a great ability to compete with oaks at the juvenile stage (Petre et al., 2012).In Romania C. betulus occupies around 4.8% of the total forest cover (IFN, 2012).This species is spread between 100 and 450 (500) m altitude, being rare in mixed beech-fir-spruce mountain forests (Horeanu, 1996).C. betulus is considered as a valuable species due to its strong ecological adaptability and increased tolerance to environmental stress, especially in extreme conditions (Șofletea and Curtu, 2007).C. orientalis is a major pioneer species in arid regions where it can colonise open and degraded areas (Bergmeier and Dimopoulos, 2008;Čarni et al., 2009).Both species cooccur in southeastern Europe and no natural hybrids between C. betulus and C. orientalis have been reported; however, controlled crosses indicated that hybrids had inferior survival and growth characteristics, and that adult hybrids were sexually immature (Santamour, 1995;Grivet and Petit, 2003).The northern edge of C. orientalis distribution range is the southern part of Romania, but isolated small populations are also present in the northern part of the Moldova region (Şofletea and Curtu, 2007).Most of C. orientalis populations in Romania can be considered as marginal populations and their genetic diversity assessment is of great importance, due to ongoing global warming.

Materials and Methods
Sampling and DNA isolation Shoots and leaves were collected from 18 populations of C. betulus and six populations of C. orientalis covering the species distribution range in Romania (Table 1, Fig. 1).The material was stored in an ultrafreezer at temperatures of -60 °C until used for DNA extraction.
Genomic DNA was extracted from biological specimens using the CTAB method described by Doyle and Doyle (1987) with minor modifications.The DNA was quantified using NanoDrop 8000 (Thermo Scientific, Wilmington, USA, 2008) and extracted DNA was stored at -20 °C.

Chloroplast microsatellites analysis
Due to the lack of intraspecific variation observed in Carpinus spp.we initially tested ten universal chloroplast microsatellites (ccmp1, ccmp2, ccmp3, ccmp4, ccmp5, ccmp6, ccmp7, ccmp8, ccmp9 and ccmp10) developed by Weising and Gardener (1999).Nine out of ten primers showed the amplification products, but only three primers (ccmp4, ccmp7 and ccmp10) showed fragment length polymorphism.DNA was diluted (1:30) prior to PCR amplification.PCR reactions were performed in a total volume of 15 µl of a reaction mixture consisting of 1× PCR Buffer (Promega), 2.5 mM MgCl2, 200 µM of each of the four dNTPs, 1 unit of Taq DNA Polymerase (Promega), 0.2 µM of each primer added to 2 µl of genomic DNA.PCR amplifications were performed in a Corbett Palm-Cycler CG1-96 with the following conditions: an initial denaturation for 15 minutes at 95 °C followed by 30/35 cycles of denaturation for 1 minute at 94 °C, annealing for 1 minute at 50 °C, extension for 1 minute at 72 °C and a final extension for 20 minutes at 72 °C (30 cycles for ccmp4 and 35 cycles for ccmp7 and ccmp10).The amplification of chloroplast SSRs was performed using fluorescence dyed forward primers (Metabion) for genotyping purpose, namely 6-FAM/Blue (ccmp7 and ccmp10) and HEX/Green (ccmp4).
Genetic variation of C. betulus and C. orientalis have been investigated with different DNA markers targeting different polymorphic regions of the chloroplast genome by PCR-RFLP (Restriction Fragment Length Polymorphism) (Postolache et al., 2017) and also by using chloroplast microsatellites (Simple Sequence Repeats; cpSSRs) (Grivet and Petit, 2003;Fărcaş et al., 2006).The chloroplast DNA diversity for C. betulus was significantly lower in western Europe compared with southeastern European populations, that harbour nearly all genetic variation and consequently a more detailed analyses in this region is absolutely necessary to quantify the genetic structure and diversity.Beside this, previously published results Petit and Grivet (2003) indicate for very low introgression between these two species.The C. betulus populations from Romania have a very distinct Holocene postglacial history with direct consequences on current genetic pattern that needs to be investigated further in order to understand the species diversity and evolutionary history (Grivet and Petit, 2003;Fărcaş et al., 2006).Moreover, the phylogeographic data on C. orientalis are very scarce and these data are very important in the conservation and management of this drought-tolerant woody species.
The aim of this study was to analyse the chloroplast DNA variation in two sister species of hornbeam (C.betulus and C. orientalis) in Romania (Northern Balkan Peninsula) using cpSSRs markers, that can bring detailed information on species genetic structure, which could be applied in developing site-oriented conservation strategy and sustainable management of Carpinus spp.genetic resources.The major objective of this work is to characterize the genetic structure of Carpinus spp. in Romania and to test weather Carpathian Mountains acted as a major geographic barrier during the Holocene postglacial recolonization history.More specifically we aimed: i) to assess the chloroplast DNA genetic variation within and among Carpinus spp.populations using cpSSRs; ii) to analyse the geographic distribution of chloroplast haplotypes and infer direct consequences on the management and conservation of Carpinus spp.genetic resources.Amplified fragments were analysed on a GenomeLab GeXP Genetic Analysis System (Beckman Coulter) using an internal size standard (CEQ 400, Beckman Coulter).Polymorphisms were defined by different length variants of detected microsatellite fragments.

Data analysis
The programme PERMUT cpSSR (Pons and Petit, 1996) was used to calculate the parameters of chloroplast DNA diversity (hS = haplotype diversity within populations based on unordered haplotypes, h T = total haplotypie diversity statistics based on unordered alleles, G ST = differentiation among populations based on unordered alleles and R ST = differentiation among populations based on ordered alleles).
A hierarchical analysis of molecular variance (AMOVA) was conducted to estimate the distribution of genetic variation among and within populations using GenAlEx version 6.5 (Peakall and Smouse, 2012).The variance components were tested statistically by non-parametric randomization tests using 999 permutations.Genetic distances (Nei, 1987) between pairs of populations were calculated with HAPLOTYPE ANALYSIS© version 1.05 (Eliades and Eliades, 2009).
The software SPAGEDI version 1.5 (Hardy and Vekemans, 2002) was used to test for the presence of a phylogeographic pattern by permuting allele sizes among alleles.If the observed R ST is higher than the R ST after permutations, we may infer that there is a phylogeographic pattern.
The relationships among found cpSSR haplotypes were inferred based on the median-joining network algorithm (Bandelt et al., 1999).Maximum-parsimony analysis was conducted using the software NETWORK version 5.0 (available at http://www.fluxus engineering.com/sharenet.htm).NETWORK identifies the number of mutational steps that separate a given set of haplotypes and accounts for haplotype frequency at a given node.
To evaluate the population phylogenetic relationships among populations an unweighted pair group average network (UPGMA) was constructed using Euclidean paired group algorithm of PAST software (Hammer et al., 2001).One thousand bootstrapped replicated matrices of pairwise Nei's genetic distances (1987) were calculated using the same program.

Chloroplast DNA haplotypes
Following the genotyping of the 128 individuals with three cpSSRs, 8 variants were identified.Thus, only two variants (and thus haplotypes) were found in the two species, which are completely different in this analysis.
One of the three chloroplast microsatellite regions showed polymorphism in C. betulus, that displayed two size variants at ccmp7 (Table 2).The size variants from all three analyzed chloroplast microsatellites were combined into two haplotypes for C. betulus, with H1 being found in all eighteen populations, while H2 was found in seven populations (Fig. 1.A).
In C. orientalis two size variants were observed at ccmp10 among three analyzed microsatellite regions (Table 2).By combining size variants of chloroplast SSRs, two haplotypes were identified in six populations.Five populations of C. orientalis from southern Romania are monomorphic (haplotype H3, while haplotype H4 was detected only in an isolated population located in north-eastern part of Romania (Fig. 1.B).
The analysis of molecular variance (AMOVA) for C. betulus showed that 42% of the haplotypic variation was due to differences among populations, whereas 58% to differences within populations.In C. betulus only 11 out of 18 populations are represented by one haplotype and 7 populations are mixed by two haplotypes with three populations significantly dominated by H2 (MM, BZC and BZGT) (Fig. 1).
The permutation test performed in SPAGEDI showed that the observed R ST was equal to the R ST after permutations, indicating an absence of a phylogeographic pattern for both species.Indeed, the dominance of two haplotypes in the Carpathians was also suggested by Postolache et al. (2017), based on the polymorphic bands observed in two combinations (atpH-atpI/HinfI and trnC-trnD/TaqI).However, a geographic pattern of genetic variation may be observed from the haplotype frequency map in C. betulus and C. orientalis (Fig. 1).Because our study found only two variants in each of the two species, the results should be interpreted with caution.

Phylogenetic relationships between haplotypes
The genetic relationships between all haplotypes show a clear (Fig. 2) separation of haplotypes specific for C. betulus and those identified in C. orientalis.Within the network, H1 is the ancestral one and H2 appears to be derived from H1.On the other hand, H3 detected in C. orientalis most probably evolved from H4, however these results need to be interpreted with caution due to limited variants found.
The UPGMA dendrogram based on Nei's distances (1987) supports geographic grouping of haplotypes by species (Fig. 3).All C. betulus populations were clustered together with subsequent subclustering of populations in two groups, each group being specific for one of the two haplotypes.On the other hand, C. orientalis populations are clustered into two distinct subclusters, with one of subcluster represented by population Roşcani, in which a unique haplotype was observed for C. orientalis.It is also important to note that branches of the major clusters correspond to distinct geographic regions.1.000 (NC) hT = total genetic diversity, hS = genetic diversity within populations, GST and RST = genetic differentiation, SE -standard errors, NC = non-significant

Discussion
Our data confirmed previous results that the two Carpinus species do not share common chloroplast haplotypes (Grivet and Petit, 2003).The non-overlapping and non-sharing of haplotypes between C. betulus and C. orientalis suggest that a much more ancient separation of these two taxa occurred perhaps even before quaternary glaciations, as discussed for other related species (Hewitt, 1999).Indeed, sequences of the nuclear ribosomal internal transcribed spacer (ITS) suggest an extensive biogeographic track between southeastern Europe and eastern Asia and most importantly indicate for two distinct clades of Carpinus where C. betulus is grouped in a different clade than C. orientalis (Yoo and Wen, 2002).This result suggests that these two species have their origin in eastern Asia considered a refugium for the section Carpinus.Paleobotanic fossils of the section Carpinus have been reported for Europe from the Middle Eocene Eckfeld flora (Wilde and Frankenhäuser, 1998).Additionally, each species has a different ploidy level with 2n = 8x = 64 for C. betulus (Petrova, 2006) and 2n = 2x = 16 in C. orientalis (Santamour, 1995).The differentiation among C. orientalis populations was very high and reached a value of which is due to the fact that populations are fixed for one haplotype or another.Our data indicates fixation of haplotype H3, which was found in five out of the six populations.The high differentiation detected in C. orientalis populations can be explained by strong fragmentation of populations that are isolated over long distances.High values of genetic differentiation was also observed in Corylus avellana (GST = 0.89) with low dispersal abilities and which is member of the same family Betulaceae (Petit et al., 2003;Leinemann et al., 2013).The low values of differentiation similar to those observed in our study for C. betulus populations were reported for Castanea sativa (GST = 0.43 - Fineschi et al., 2000) and Betula pendula (GST = 0.424 -Palmé, 2003;Leinemann et al., 2013).It was estimated that, on average, the woody angiosperms present a differentiation index between populations of 0.73 (Petit, 1999).The relatively low genetic differentiation among C. betulus populations may be explained by historical effective seed dispersal over short but also over long distances during postglacial recolonisation (Heuertz et al., 2004).
The geographic distribution of cpDNA haplotypes in C. betulus reveals a phylogeographic pattern, which is comparable with a recent study on hornbeam by Postolache et al. (2017) based on PCR-RFLP analysis.Haplotype H2 is preponderant in populations located on the external side of the southeastern Carpathians arch (Gura Teghii -75% and Colţi -80%).In this area, palynological analysis revealed early records of C. betulus at lower altitudes in the Subcarpathian region (Magyari, 2002;Willner et al., 2009).The same haplotype H2 was also observed in a quite distant C. betulus population located in the north-west part of Romania (Apa Sărată), the Maramureş region, where it has a relative frequency of 60%.It is worth to mention the high frequency of H2 in the population (MM) from northern part of Romania and in two populations from the Curvature Carpathians (BZC and BZGT).These populations are very close to the 'suture zones' identified in the Ukrainian and in the Curvature Carpathians, which resulted as a consequence of contact between different recolonization routes (Postolache et al., 2017).Delineated postglacial recolonisation routes by Postolache et al. (2017), confirm our phylogeographic pattern and explain the observed geographic distribution of haplotypes (H1 and H2).
The existence of a northern latitude refugium for C. betulus was previously mentioned in the seminal work of Willis (1996).Indeed, this was also suggested by Magyari (2002) who mentioned the existence of a glacial refugium for C. betulus in the North Hungarian Middle Mountains, based on palynological data.More recently, Mitka et al. (2014) also indicated a possible refugium in the Northern Carpathians, and raised the question of whether post-glacial recolonisation would have been plausible from this northern refugium.However, based on the recent findings (Postolache et al., 2017), we can suggest that haplotype H2 most probably originated in the south-eastern part of Bulgaria.The glacial refugium for haplotype H1 was most probably located in the mountains of Rhodope and Pirin, while the glacial refugium for H2 was located separately in the Strandzha Mountains, very close to the Black Sea coast.The post-glacial re-colonization routes B and C that correspond to haplotypes H1 and H2, respectively, were described in detail by Postolache et al. (2017).
The two haplotypes detected in C. orientalis are located in two distinct geographic areas: haplotype 3 being specific for southern part of Romania, while haplotype 4 was observed in an isolated population located at the northern limit of species range.Haplotype H3 is more present in the area of forested steppe lands, located in the southern part of the Carpathians and may suggest that those populations might have a common origin most probably as a result of recolonisation from southern Balkan refugia (Grivet and Petit, 2003).The presence of a distinct haplotype (H4) in the Roşcani population may be explained by a different glacial refugium origin.Based on this finding we may suggest that Roşcani population deserve high priority for conservation and can be included in the genetic resources of this species.Interestingly, haplotype H4 that was identified 200 kilometres farther north than the nearest southern C. orientalis populations (Jideni, Ramnicu Sărat).Isolated C. orientalis populations are also found in Vaslui (Bogdana, Lupesti, Stănileşti) (Chifu et al., 2006), at a distance of 100-110 km from Roşcani population, therefore this population would deserve more attention in future studies with both chloroplast and nuclear SSRs.Roşcani population that harbour haplotype H4 consists of a small number of trees, that was possibly larger in the past although there is no evidence of recent bottleneck (based on nSSRs analysis (data not shown).The origin of this unique haplotype is difficult to be explained, but it is worth to mention that this C. orientalis population is located at the limit of species natural distribution range in highly fragmented habitats.
Previous studies have also shown that C. betulus populations from Romania harbour outstanding genetic polymorphism compared with western European populations (Grivet and Petit, 2003).Indeed, we observed seven polymorphic C. betulus populations, out of 18, using only three cpSSR markers.All six populations of C. orientalis are fixed for one of the two detected haplotype.To our knowledge, this is the first study on C. orientalis genetic structure of populations from Romania.The small number of observed haplotypes in both species may suggest that only few glacial refugia of Carpinus species have existed in southern Balkans during the Late Glacial Maximum (LGM).We may thus expect that populations of Carpinus species from southern Balkans may harbour unique haplotypes, which was confirmed in a recent study (Postolache et al., 2017), but those populations did not succeed to expand and to migrate because of tough competition with early successional trees species (Grivet and Petit, 2003).Additionally, the high heterogeneity of mountain landscape across Balkan Peninsula may hinder expansion northwards of Carpinus species.Reduced genetic chloroplast DNA variation was also observed in beech (Demesure et al., 1996), which is also considered a latesuccessional tree species and also in hazel (Palmé, 2002;Leinemann et al., 2013).Most of these studies suggest that long glacial episodes have a profound effect on geographic organization of chloroplast DNA variation.The analysis of chloroplast genome in two Carpinus species (C.betulus and C. orientalis) revealed a distinct geographic distribution of cpDNA haplotypes, which emphasize that hornbeam forests have undergone idiosyncratic postglacial migration.The interspecific phylogeography of Carpinus spp.succeeded to reveal hornbeam populations with peculiar genetic structure (e.g.Roşcani), which may have important consequences on the management and conservation of Carpinus spp.genetic resources in Romania.

Conclusions
Only a limited number of chloroplast haplotypes was observed in both Carpinus species across Romania.The geographic distribution of chloroplast haplotypes suggests that Carpathian Mountains may have played an important role during the Holocene postglacial recolonization history of Carpinus species.Our study confirms that both Carpinus species do not share common haplotypes, which may suggest a much more ancient separation of these two taxa.A very rare chloroplast DNA haplotype was identified in an isolated peripheral population of C. orientalis, which deserve a high priority for conservation.The geographic structure of chloroplast genetic variation in Carpinus species appears to be primarily a reflection of past (postglacial) migration history, whereas more recent adaptive and stochastic processes have occurred at more local scales.

Fig. 2 .
Fig. 2. Maximum parsimony phylogenetic network among four chloroplast haplotypes observed in three chloroplast SSRs loci.Size of circles corresponds to number of individuals of a given haplotype

Table 1 .
Geographic location of populations

Table 2 .
Observed chloroplast DNA haplotypes in C. betulus and C. orientalis

Table 3 .
Levels of genetic diversity and differentiation in Carpinus spp.