Genetic melting pot and importance of long-distance dispersal indicated in the Gladiolus imbricatus L. populations in the Polish Carpathians

The genetic diversity in 11 populations of Gladiolus imbricatus in five mountain ranges, including the Tatra, Pieniny, Gorce, Beskid Niski (Western Carpathians) and Bieszczady Mts (Eastern Carpathians), was studied with inter-simple sequence repeat (ISSR) markers. The species is a perennial plant occurring in open and semi-open sites of anthropogenic origin (meadows and forest margins). We checked a hypothesis on the microrefugial character of the plant populations in the Pieniny Mts, a small calcareous Carpathian range of complicated relief that has never been glaciated. Plant populations in the Tatra and Pieniny Mts had the highest genetic diversity indices, pointing to their long-term persistence. The refugial vs. the non-refugial mountain ranges accounted for a relatively high value of total genetic variation [analysis of molecular variance (AMOVA), 14.12%, p = 0.003]. One of the Pieniny populations was of hybridogenous origin and shared genetic stock with the Tatra population, indicating there is a local genetic melting pot. A weak genetic structuring of populations among particular regions was found (AMOVA, 4.5%, p > 0.05). This could be an effect of the frequent short-distance and sporadic long-distance gene flow. The dispersal of diaspores between the remote populations in the Western Carpathians and Eastern Carpathians could be affected by the historical transportation of flocks of sheep from the Tatra to Bieszczady Mts.


Materials and methods
Study species. The genus Gladiolus L. belongs to the Iridaceae family and has two distinct areas of geographic distribution. The main center is in Eastern and Southern Africa (Capensis), where more than 100 species have been described. The second area of distribution is in the southern part of Eurasia, especially in the Middle East and the Mediterranean region. In this area, the number of species is much smaller. Hamilton 17 listed six species as indigenous to Europe. European species of Gladiolus are polyploids with a chromosome number from 60 in tetraploids to 180 in dodecaploids, but aneuploids there are known to exist. The evolution of the European Gladiolus species has been affected by hybridization and polyploidy. Gladiolus imbricatus covers Central and Eastern Europe, the Mediterranean, Caucasia, and West Siberia 13,17 .
G. imbricatus is an anemochore, perennial species, and a bulbo-tuber geophyte. It blossoms in July, pollinated by Lepidoptera. It occurs throughout Poland, but it is a rare species, more often inhabiting mountainous areas. Available historical data suggest that it was common in southern Poland in the nineteenth century [18][19][20][21] . It occurred in thermophilous oak forests of Potentillo albae-Quercetum, in moist lowland meadows of the Molinion alliance 22,23 , and as a weed in oat and barley fields 10,24 , probably as a remnant of previous meadow communities 18 . It was also a co-dominant species in many types of upland pastures and meadows 10,15,18 . This state has persisted until modern times; there has only been a drastic decline in the number of its sites in the last few decades. This process is particularly visible in the central and northern parts of the country, where the appearance and disappearance of new populations is observed 25,26 . In southern Poland, population declines and fragmentation are also recorded. As a result, G. imbricatus is now endangered in Poland 27 . Moreover, it is gradually disappearing in the whole of Europe 13,23,28 . In Poland, the plant species is under law protection. Sample collection. Plant material (fragments of ca. 4 cm 2 ) to DNA isolation was collected in nine populations in the Western Carpathians and in two populations in Eastern Carpathians (Bieszczady Mts, Table 1). Populations TKG, TKD, and BW represent moderately dry variant of Pieniny's meadow Gladiolo-Agrostietum capillaris; JU-nitrophilous dry meadow with the domination of Linaria vulgaris; LU, LB (Gorce), LP (Tatra) S, B (Bieszczady), BN (Beskid Niski Mts) represent wet meadows of Molinion caeruleae alliance ( Fig. 1 DNA isolation and ISSR analysis. DNA was isolated from fully developed fragments of leaves (4 cm 2 ) without damage symptoms caused by insects and mold. DNA was extracted with Genomic Mini AX Plant (A&A Biotechnology) according to the manufacturer's instructions. The sequences of primers were taken from Stepansky et. al. 29 and are shown in Table 2. PCR reactions were conducted with a 2720 Thermal Cycler (Applied Biosystems). The composition of the reaction mixture and the PCR reaction conditions according to Gupta et al. 30 and optimized by Sutkowska 31  www.nature.com/scientificreports/ Products were subjected to electrophoresis on 1.5% agarose gel stained with ethidium bromide (50 µl/100 ml) at 100 V about 1.5 h. PCR products (bands patterns) were observed and archived with Imagemaster VDS (Pharmacia-Amersham).
For analysis of the length of PCR products, GelScan ver. 1.45 (Kucharczyk TE) software was used. ISSR reproducibility tests included within-plate (n = 12) and between-plate (n = 9) replicates independently analyzed from the DNA extracts 32 . Data analysis. All the analyses performed were based on the following assumptions: ISSR markers behaved as dominant markers; co-migrating fragments were considered homologous loci; and populations were at Hardy-Weinberg equilibrium, in which case allele frequencies were estimated from Bayesian method with the non-uniform prior distribution of allele frequencies, to calculate expected heterozygosity Hj. Statistics of genetic diversity and population genetic structure were computed after estimating allele frequencies, including the percentage of polymorphic bands (PLP) at 95% criterion. The calculations were performed with AFLP-SURV version 1.0 33 .
Shannon's diversity index (I) was calculated to provide a relative estimate of the degree of genetic variation within each population, based on the formula I = -Σ Pi log2 Pi, where Pi was the frequency of each ISSR band. The number of alleles (bands) with nonzero frequency (allele number na) and the reciprocal of homozygosity (effective allele number ae) were calculated using POPGENE version 1.32 34 . As an additional diversity marker, the rarity index DW, corresponding to "frequency-down-weighted marker values" per individual 35 was computed using AFLPdat 36 . High DW values are expected in long-term isolated populations 37 . Analysis of molecular variance (AMOVA) 38 was carried out using the program Arlequin version 3.11 39,40 .
Matrix of Nei's genetic distance 40 was calculated for each pair of 11 populations with AFLP-SURV 33 . The matrix was then used to construct an NJ tree 41 with reticulations with the T-REX ver. 4.1 software 42 , and the stopping rule for reticulations based on Criterion Q1 43 . The bootstraps were calculated in CONSENSE option implemented in PHYLIP version 3.6 44 based on 1000 random runs.
The matrix of 77 samples × 167 ISSR products coded as binary (0-1) data was used to calculate Nei and Li (Dice) distances 45 . They were used for a non-metric multidimensional scaling analysis (NMDS) 46 performed with the NTSYSpc ver. 2.11 multivariate analysis package 47 . NMDS displays a monotone relationship to the distances implied by the original data matrix and achieves a much better fit in fewer dimensions than is possible with other ordination methods.
The genetic division among individuals in populations of both species was estimated using STRU CTU RE, version 2.3.3 48 , applying a Bayesian model-based clustering algorithm for the use of dominant markers 49 . The numbers of K = 1-5 groups were tested in ten replications per K. A burn-in period 100 000 was applied, followed by a procedure using 500 000 Markov chain Monte Carlo (MCMC) repetitions. The estimation of the optimal number of groups was based on the likelihood of partitions, estimates of posterior probability provided in STRU www.nature.com/scientificreports/ CTU RE output, examined as a function of increasing K 48 and ΔK values, estimating the change in the likelihood function with respect to K and estimated as an indicator of the most reliable clustering structure 50 . The similarity between runs was estimated using the symmetric similarity coefficient 51 with the R-script Structure-sum-2011 36 . K = 2 represented an optimal clustering structure.

Results
In the ISSR analysis of 77 individuals with 4 primer combinations, 165 unambiguous fragments (loci) were selected ranging between 130 and 990 bp. The number of PCR products per individual varied in particular primers within 3-13 ( Table 2). The within-population genetic variation was calculated using different indices (Table 3). All genetic variation values were consistently highest in populations Tatra JU and Pieniny BW. The mean allele number na and the effective number of alleles ne ranged between 1.077-1.440, and 1.078-1.192, respectively. The mean percentage of polymorphic loci (PLP) ranged between 22.6 and 44.0%, and Shannon's I index was between 0.049 and 0.187. Higher values for genetic variation were recorded in the Tatra Mts, then in the Pieniny Mts. Lower values were found in the Gorce and Bieszczady Mts. The survey of the rarity index DW revealed a similar pattern of variability across regions. Its highest values were found in Tatra JU and Pieniny BW, and the lowest in Beskid Niski BN and Gorce LB.
The first group from the Pieniny BW formed individuals of the hybridogenous origin, according to the results of the Bayesian STRU CTU RE analysis. Also, a hybridogenous is an individual Pieniny TKD8 (STRU Table 2. The primers used in PCR, primer sequence, total number of reactions products generated by each primer and range fo number of PCR product per specimens.   Fig. 2A). All hybrid individuals have genetic admixture from the population Tatra JU. The populations Pieniny BW and Tatra JU formed a highly (91%) supported sister group (NJ, Fig. 2B). The second group of Tatra JU is genetically heterogenous: four individuals show the links with the population Pieniny TKD and, to a lesser extent, with Bieszczady S. The third group forms the remaining populations. Among them, the fairly uniform Tatra-Gorce LP/LU group is visible, with high support of 80% (NJ). The second population Gorce LB is presumably of the hybridogenous origin, with the admixture of Tatra JU (STRU CTU RE-result not shown, Fig. 2A). www.nature.com/scientificreports/ The results of the NJ classification with reticulations show the existence of the three clusters (Fig. 2B). The first group, with a moderate support of 64%, consists of Pieniny BW and Tatra JU (highly supported with 91%), Bieszczady B, S, and Gorce LB. Here a reticulation between Pieniny BW and Bieszczady B was found. The second group forms an unresolved cluster with Gorce LU and Tatra LP (80% support). The third group consists of the remaining Pieniny populations TKD, SN, TKG (50% support) and the population from Beskid Niski BN (78% support, Fig. 3).
A reticulation between Gorce LB and Gorce LU was noted. Bayesian analysis (STRU CTU RE, K = 2) show the distribution of genetic diversity among populations (Fig. 3). The genetically most specific was population Tatra JU. Population Pieniny BW, and in a lesser extent Gorce LU were hybridogenous. Two remote Bieszczady's populations B and S had a small admixture of the genetic diversity from Tatra JU/Pieniny BW/Gorce LU (Fig. 3).
The above genetic links between geographically remote populations are supported by the occurrence of specific bands (alleles), occurred exclusively in two or three geographical regions. The populations from Bieszczady Mts B shared two unique bands with Pieniny BW and two unique bands (Bieszczady S) with Pieniny TKD. The NJ group Gorce LU/Tatra LP shared two unique bands with Beskid Niski BN. The only population with two private bands was Pieniny TKD (Table 3).
Non-hierarchical AMOVA showed a significant but moderate level of genetic divergence among 11 populations of G. imbricatus: 37.61% (p < 0.001). The analysis of the genetic structure of the populations in five geographical regions showed small, statistically insignificant, population differentiation in the particular mountain ranges (4.50%, p > 0.05). The test a hypothesis on the genetic divergence of the populations in two presumably refugial ranges: Pieniny and Tatra Mts gave the statistically significant result: 14.12% (p = 0.003) ( Table 4).

Discussion
Open-site natural vegetation forms peculiar plant communities in high-mountain zones (alpine vegetation), northern latitudes (arctic tundra belt), and continental climates (steppe formations). Historically, it also occurred in the Carpathian forelands and adjacent lowlands in the Late Vistulian (the Oldest Dryas), forming the tundra and cold steppe formations, according to many pollen spectra. No pollen of Iridaceae was found at that time in Poland 52 . Our results show that G. imbricatus could have persisted locally in the Last Glacial Maximum in the Tatra Mts and Pieniny Mts (Western Carpathians). The latter calcareous mountain range could have formed a local refugium and genetic melting pot for many plant species.  54 , and amounted to 38%, and 0.130, respectively. The reason for such high genetic diversity is unknown; the population was perhaps related to an unrecognized "cryptic" refugium nearby.
Generally, the higher genetic diversity of Carpathian populations compared to those in lowland areas could be an effect of their refugial character (especially in the Pieniny and Tatra Mts). For AMOVA, these two geographical regions accounted for 14% (p = 0.003) of the total genetic variance of the Carpathian populations of G. imbricatus. The result showed the statistically significant genetic divergence between refugial and non-refugial areas, with the outcome being especially convincing when compared with the non-significant structuring of the populations in five geographic regions of the Polish Carpathians. Populations of the species localized more to the north in Poland seem genetically impoverished in comparison with those occurring in the mountains. Similar results concerning the lowland vs. mountain populations showed studies on forest grass Bromus benekenii 55 and wet meadow Primula farinosa 56 , which could have originated from the glacial refugia, especially those located in the Carpathian refugium 57 .
A much higher among-population component of diversity (AMOVA) that amounted to 86% displayed a grass, Melica transsilvanica, studied in the Pieniny Mts and Małopolska Upland (southern Poland) 58 . The species has a different ecological profile than G. imbricatus; it grows on dry, exposed outcrops in pioneer rupicolous grasslands (Seslerio-Festucion duriusculae alliance in the Pieniny Mts) and in xerothermic grasslands of the Cirsio-Brachypodion pinnati alliance on the Małopolska Upland. The high level of among-population diversity was linked with a decreased within-population diversity index (PLP 3.7%) and nearly total genetic homogeneity. Taken together, the results of the studies on M. transsylvanica in the Pieniny Mts point to the high geographical isolation of the metapopulation in the region and the dominant self-pollinating system of the species is inducing the lack of gene flow 58 .
The relatively high within-population genetic diversity of the G. imbricatus populations in the Carpathians could be related to the presumably refugial character of some Western Carpathian populations. This has been proven by the occurrence of the private bands, a typical phenomenon in the relict areas 59 , and the high value of the DW index 37,60 . Accordingly, the high genetic diversity could be the effect of the long-time presence of large population numbers in the interglacial 15 , counteracting the effects of genetic drift or genetic bottleneck that affect smaller populations more heavily. An example is the extant Polish population of Primula farinosa in the Pieniny Mts which probably formed a continuous distribution in the interglacial between the Carpathian Mts and Polish lowlands 56 . Today, the relict plant population demonstrates a distinct genetic makeup and admixture of various genetic lineages, an example of the Pieniny genetic melting pot. Increased diversity was obtained through the redistribution of genetic information in a given place by dispersal from refugial areas 55,59,60 . Our study of G. imbricatus also provides clear phylogeographic evidence on the existence of a local genetic melting pot in the Pieniny Mts. The Bayesian inference shows the two genetic types of the species in the Polish Carpathians. They formed a hybridogenous population Pieniny BW, the sister group, with a donor population Tatra JU. In the other population, Pieniny TKD, one individual (TKD8) also shows the genetic admixture from Tatra JU. It is compelling evidence of the local short-distance migrations. Moreover, frequent short-distance dispersal could probably prevent the effects of genetic drift (genetic bottleneck) in the distracted and divided populations of southern Poland. www.nature.com/scientificreports/ The remaining populations in the Pieniny Mts formed a distinct but marginally supported cluster (50%), showing the different histories of their origin. The same concerns apply to two populations in the Gorce Mts. Geographically adjacent, they have not originated from the same genetic stock; the first (Gorce LU) has close genetic links with Tatra LP (NJ), and the second (Gorce LB) with Tatra JU (STRU CTU RE, NMDS). Interestingly, the most remote populations, Bieszczady B and S, were linked genetically to the Pieniny BW population, as well as with Gorce LU and Tatra LP (a distance of ca. 250 km). The reticulation in NJ and the sharing of the rare specific alleles in such remote populations as Bieszczady B and S, and Gorce/Tatra group LU/LP, together with Pieniny BW, TKD, SN, point to the occurrence of long-distance dispersal. The placement of the population from Beskid Niski BN in one group with Pieniny populations could characterize the stochastic process.
The genetic similarity of the distant populations (Bieszczady vs. Tatra Mts) may be a result of numerous overlapping factors, including multidirectional gene flow in the dispersal history, long-distance dispersal during postglacial recolonization, and survival in several detached refugia 61 . All of these factors led to diminishing the among-population diversity component. We found rare, shared bands in remote populations that can most likely be explained by long-distance migration. The relevant example is the close genetic links between the Pieniny/ Tatry and Bieszczady Mts. The long-distance migration in G. imbricatus, which has limited dispersal ability 62 , could be explained by human impact. For example, the transport of numerous sheep flocks (up to 50,000 individuals) by rail from the Tatra Mts to the Bieszczady Mts in the 1960s was the result of overgrazing in the Tatra Mts area, and the administrative decision of the Tatra National Park authorities 64 . Also, many villages in the Beskid Niski Mts were abandoned after the Second World War and the flocks of sheep from the Tatra/Pieniny regions were then introduced there. In our opinion, this could be a crucial factor in determining the close genetic relations between populations in such remote regions.
The similarities among populations in close areas of the Pieniny, Tatra, and Gorce Mts, where sheep flocks were abundant, densely distributed, and interchanged often in historical times, could be explained by shortdistance diaspore dispersal.
While studying the genetic diversity of Aconitum species in the Carpathians and the Sudetes 63 , an introgression was found between species with non-overlapping geographical ranges, probably as a result of seed dispersal with pasturing within a distance of ca. 20 km (Mantel test). It seems to be an average of short-distance seed dispersal in the mountains, in the realm of the traditional agro-pastoral economy.
The lack of genetic differences between populations in different regions is a common phenomenon in phylogeographic studies. For example, in a study on the genetic diversity of a subalpine perennial Cicerbita alpina across Europe 65 , the authors found a weak phylogenetic structure of the tall-herb community species among Carpathians regions (Eastern Carpathians, Southern Carpathians, Western Carpathians) (AMOVA among-group component equaled 9.78%). Also, the level of genetic divergence of the species (30%) in the non-hierarchical analysis is lower than an average in the case of mountain species. The result was explained by the biological characteristics of the species: a tall plant (up to 2.5 m), producing a large number of small diaspores over long distances, can promote gene flow between geographically distant regions and result in low parameters of genetic isolation.
Summing up, glacial refugia of G. imbricatus could have existed in the Western Carpathians. An example is the population Tatra JU, the most genetically diversified population forming a distinct genetic group among the populations that have been studied. A hybridogenous population was found in the Pieniny Mts, which are generally believed to be a glacial refugium, proving the dispersal from the Tatra Mts and the existence of the local genetic melting pot. The other dispersal from the Tatra Mts refugium was also traced adjacent to the Gorce Mts. The unexpected genetic links found between the remote populations of G. imbricatus, namely the Pieniny/ Tatra and Bieszczady populations, could be explained by long-distance dispersal linked with the transportation of sheep flocks between the Tatra and Bieszczady Mts in the second half of the previous century. The migrations of flocks were accompanied by the plants (diaspores), which facilitated the mixing of hitherto isolated populations. In effect, a lowering of the values of genetic divergence between distant populations of G. imbricatus was noted. In this way, the genetic links between plant populations in close or distant mountain ranges could be affected by the recent, stochastic processes leading to the decreasing values of genetic divergence indices. In conclusion, the role of many centuries of agro-pastoral human economic activity in the migration of the plant species seems, in some cases, crucial.

Data availability
Data on genetic variation in Gladiolus imbricatus are available from the authors upon request. www.nature.com/scientificreports/