The genetic diversity of marginal populations is necessary for adaptation to local conditions and, at the same time, for the ability to expand the range. The narrowing of the amplitude of ecotopes available for habitation at the edge of the range potentially strengthens stabilizing selection, reducing variability and, accordingly, the potential for adaptation to new ecological niches. However, the weakening of the gene flow from the center of the range decreases the supply of genes that are maladaptive to local conditions and improves the possibilities for expansion [1].

There are three pine species naturally growing in the Crimea and Caucasus: one boreal species, Scots pine (Pinus sylvestris L.), and two Mediterranean species, Calabrian pine (P. brutia Ten.) and European black pine (P. nigra Arn.) [2, 3], which are isolated from their main ranges. Scots pine forms an isolated enclave in the Crimea and the Caucasus: due to severe competition with broad-leaved forest species, it has been displaced to small isolated rocky areas that are unsuitable for its competitors. In turn, Mediterranean pines P. brutia and P. nigra in the Crimea and the Caucasian coast of the Black Sea are at the limit of resistance to low winter temperatures [4, 5].

Pinus brutia is particularly prominent in this context: it is an East Mediterranean species with the range covering mainly Asia Minor [2, 3, 6, 7]. Its stands in the Crimea and Caucasus (the extreme northeastern limit of P. brutia natural distribution) are considered the most ancient relict formations of pine forests [8]. This pine occurs only within a few hundred meters from the sea, which smoothes the peaks of lowest winter temperatures. In the Crimea, the range of P. brutia includes only the southern part of the peninsula, where it grows in two areas separated from each another: at Cape Aya (about 440 ha near the city of Balaklava) and 120 km to the east, near the city of Sudak (about 20 ha). Thus, the Crimean P. brutia populations include about 15 000 trees, and their total range does not exceed 460 ha [7, 9]. In the Caucasus, the range of P. brutia is a narrow belt of small groves extending from the vicinity of Anapa to Cape Pitsunda [2, 5, 10, 11]. The total area occupied by its populations in Krasnodar krai is no more than 1100 ha. Pinus brutia grows at low elevations, from the sea level itself to 200–300 m a.s.l.; it often occupies limestone rocks and rarely occurs in the plain, i.e., in the cone delta of the Bzyb River (Pitsunda).

There is still no consensus on the taxonomic status of the Crimean and Caucasian populations of P. brutia [2, 6, 7, 12]. In recent times, they have been usually regarded as taxa of the same species [2, 13] or as subspecies or variations. According to Farjon [13], the Crimean–Caucasian populations belong to P. brutia var. pityusa (Steven) Silba. Despite the small size of its populations, P. brutia var. pityusa shows wide morphological variability [12], and some Russian morphologists (e.g., [14]) consider the groups of Caucasian and Crimean populations as separate species: Pitsunda pine P. pityusa Stev. in the Caucasus and Stankewiczii pine P. stankewiczii Sukacz. in the Crimea. Moreover, Orlova [14] regards the northernmost populations of the Caucasian coast as hybrid between P. stankewiczii and P.pityusa.

All previous studies on the genetic variation of the Crimean and Caucasian populations of P. brutia have been performed using allozyme analysis [15–19]. The allozyme data show that the Pitsunda and Stankewiczii pine populations do not differ at the species level and are quite close to P. brutia from the main range [15, 17]. Sannikov et al. [19] did not differentiate the two studied Crimean samples from the population of the Black Sea coast of the Caucasus. Korshikov and Gorlova [18] compared two natural populations growing in the Crimea (the Ayazma and Novy Svet stows) and revealed no significant genetic differences between them. However, previous genetic studies on the northeastern populations of P. brutia used only allozyme markers and did not cover its entire range in the Crimea and the Caucasus. In addition, the estimates of its variability varied depending on what set of loci was used: some authors characterized it as low [16], while others [15, 18] found it to be fairly high, with the variability in the Crimean–Caucasian populations being not lower than in populations from Turkey and Greece [15].

Further studies with different types of genetic markers should elucidate the genetic diversity, differentiation, and origin of relict P. brutia and P. nigra populations and reveal the presence or absence of the recent gene flow, which is particularly important for developing measures to protect endangered species and populations. Chloroplast microsatellites, or chloroplast simple sequence repeats (cpSSRs), have a high mutation rate, and are highly variable and selectively neutral [20, 21]. The chloroplast DNA (cpDNA) of the species of the family Pinaceae is paternally inherited (transferred with pollen), which provides for a high level of intraspecific gene flow and allows reliable delimitation of species [22]. Due to uniparental inheritance, the effective population size for cpDNA markers is half that for nuclear markers, making them more sensitive to population reduction [21, 23, 24]. The cpSSR loci have been used to study the population structure, gene flow, and demographic history of coniferous species [25–33].

Studies of P. brutia using cpSSR markers have previously been carried out in the main (Mediterranean) part of the range, namely, in Greece and Turkey. These markers clearly discriminated between closely related taxa of the P. halepensis complex, including P. brutia (Turkey), as well as the taxa of P. brutia subsp. eldarica (Iran), in which a decrease in population variability was revealed [25]. The cpSSR-based study on the role of elevation in the genetic differentiation of mountain populations of P. brutia in Turkey [34] showed a significant subdivision of populations growing at different elevations, which could be explained by genetic adaptability of P. brutia to local conditions.

The purpose of this research was to study the level of genetic variability and differentiation of relict P. brutia populations in the Crimea and Caucasus relative to each other and to populations from the main range, clarify their taxonomic status, and compare the pattern and level of variation in Black Sea P. brutia and the other two pine species growing in this region, P. sylvestris and P. nigra. For this purpose, we analyzed the variation of cpSSR loci in seven P. brutia populations in the Crimea, Caucasus, and Asia Minor and also in the Crimean–Caucasian populations of P. sylvestris and P. nigra.

MATERIAL AND METHODS

The polymorphism of cpDNA microsatellite loci was analyzed in seven populations of P. brutia, three Crimean–Caucasian populations of P. nigra, and three populations of P. sylvestris (Table 1, Fig. 1). Each population sample contained needles from 23–50 trees growing at a distance of at least 50 m from each other. Four samples of P. brutia were studied on the Black Sea coast of the Caucasus: from its northernmost habitats (Bolshoy Utrish) to one of its southernmost ones (Cape Pitsunda). We studied both habitats of this species in the Crimea: the Ayazma stow (50 trees) and Novy Svet stow (40 trees). Specimens from each Crimean population were collected in a number of localities at a distance of several kilometers from each other. Unlike previous studies [15–19], ours covers the entire range of P. brutia var. pityusa in the Crimea and Caucasus, and we also studied 15 individuals from the natural population of Pinus brutia var. brutia in its main range in the Mediterranean (Antalya, Turkey) (Table 1).

Table  1. Geographical location of the studied population samples and genetic diversity of chloroplast microsatellite loci (cpSSR) in three pine species
Full size table
Fig. 1.
figure 1

Distribution of cpSSR haplotypes (I–X) in seven Pinus brutia populations and the ranges and studied populations (1–13) of the three pine species in the Black Sea region. The range of P. brutia in Asia Minor is given from EUFORGEN 2009 (http://www.euforgen.org).

Full size image

Crimean Pine is an eastern subspecies of European black pine and has a Latin name of Pinus nigra subsp. pallasiana (Lamb.) Holmboe. We studied this subspecies in two populations in the Crimea and its only habitat in the Caucasus, namely, near the village of Arkhipo-Osipovka (see Fig. 1). The total sample sizes were 74 trees for P. nigra and 87 trees for P. sylvestris (two samples from the western Caucasus and one from the Crimea).

DNA was isolated from fresh needle samples using the CTAB method in accordance with the protocol [35]. We studied four cpSSR loci [27]: Pt15169, Pt26081, Pt30204, and Pt71936, for which polymorphism was previously detected in P. sylvestris from the main range [31]. In P. brutia samples, two more cpSSR loci were tested, Pt41093 and Pt36480, which were previously found to be variable in populations from Greece and Turkey [25]. The PCR, electrophoresis, and visualization of PCR products were performed as described [26, 31]. The fragment size variants were considered as alleles. The shortest fragment was assigned number 1, the fragment that was one nucleotide longer was assigned number 2, etc. Since there is no recombination in cpDNA, the combination of alleles of variable loci was considered as a haplotype. To verify the structure and size of an amplified fragment containing microsatellite, we sequenced each detected allele using a BigDye v.3.1 kit (Applied Biosystems). The resulting sequences were manually edited and aligned with BioEdit [36]. The following indices of variability were calculated for each population: total number of haplotypes (N) and unbiased haplotype diversity (H) [37].

Analysis of molecular variance (AMOVA) covered two hierarchical levels (within populations and between populations). We calculated two fixation indices: GST [37], which is based only on haplotype frequencies, and RST [38], which takes into account the genetic distance between haplotypes: \(D_{{sh}}^{2}(i,j) = {{K}^{{ - 1}}}{{\left[ {\sum {\left| {{{a}_{{ik}}} - {{a}_{{jk}}}} \right|} } \right]}^{2}},\) where aik and ajk are the numbers of repeats of the kth microsatellite of the ith and jth haplotypes, and K is the number of loci. The statistical significance of the fixation indices was estimated using 1000 permutations. Calculations were made with Arlequin v. 3.5 [39]. The RST and GST were compared using the algorithm [40] in PermutCPSSR (http://www.pierroton.inra.fr/ genetics/labo/Software/). When RST is significantly higher than GST, the genetically similar haplotypes tend to coexist in the same population. The ordination of P. brutia populations was done by means the principal coordinate analysis with NTSYS-pc [41] using the matrix of pairwise RST values. To assess the relationship between the genetic and geographic structures of P. brutia, we checked the correlations between genetic and geographic distances with Mantel test [42] in NTSYS-pc using the matrix of pairwise RST and matrix of geographical distances between the samples [41].

RESULTS

Species Specificity of Chloroplast Microsatellite Loci in Three Pine Species

All four loci (Pt15169, Pt26081, Pt30204, and Pt71936) were variable in Crimean–Caucasian samples of P. sylvestris; Pt15169, Pt71936, and Pt26081 were variable in P. brutia; and Pt30204, Pt71936, and Pt26081 were variable in P. nigra. The sequencing of allele variants of each locus confirmed the accuracy of typing and showed that the intraspecific allele variants with the most similar electrophoretic mobility differed from each other by one nucleotide according to the step-by-step model of the microsatellite mutation process.

Species specificity of variability was revealed for four loci, being expressed in the length and nucleotide sequence composition of fragments. Pinus nigra and P. brutia sharply differed from P. sylvestris and from each other in the Pt15169 locus (Fig. 2): P. nigra had one size variant (1n allele), which was seven base pairs shorter than the shortest allele in P. sylvestris, and P. brutia differed from P. nigra by the point mutation of G in the microsatellite and by the length of the fragments, the shortest of which (1b allele) was one nucleotide longer than the 1n allele in P. nigra. The fragment variants were even longer in P. sylvestris due to insertion (ATCT) in the flanking region of the microsatellite (see Fig. 2).

Fig. 2.
figure 2

The variable region of the Pt15169 locus in three Pinus species. Insertion in P. sylvestris is in bold and the point mutation in P. brutia is shown in gray. The allele variability is described in the text.

Full size image

At the Pt71936 locus, six alleles were recorded in P. sylvestris, the same number of alleles was recorded in P. brutia, and three alleles were found in P. nigra (Fig. 3). The size variants differed in the number of T repeats. At the Pt26081 locus, both P. sylvestris and P. nigra had three alleles and P. brutia had two alleles, with the number of T repeats being different. In addition, P. brutia and P. nigra had one A nucleotide after T repeats in the microsatellite sequence, which differs them from P. sylvestris (with two A nucleotides).

Fig. 3.
figure 3

Allele frequencies (%) at variable cpSSR loci in P. brutia (a), P. nigra (b), and P. sylvestris (c).

Full size image

Only one allele was found at Pt30204 locus in P. brutia, which was shorter than the shortest variant in P. sylvestris and P. nigra by three and four nucleotides, respectively; therefore, this locus is diagnostic for P. brutia. P. sylvestris had seven alleles at Pt30204 locus. P. nigra had six size variants at this locus; however, their sequences differed from those of P. sylvestris alleles with the same length: at the beginning of the microsatellite, both P. nigra and P. brutia had two T nucleotides, while P. sylvestris had three or four .

Therefore, the diagnostic locus for the three studied pine species is the Pt15169 locus, which makes it possible to determine the species identity without sequencing. The Pt30204 locus is diagnostic for P. brutia when electrophoresis is only used; in turn, sequencing also makes this locus diagnostic for P. nigra. The Pt26081 locus is diagnostic for P. sylvestris when sequencing is used; however, it does not differentiate between individuals of P. brutia and P. nigra. In addition to differences in the length and nucleotide sequences of the alleles, three pine species also significantly differed from each other in allele combinations in the haplotypes and had no common cpSSR haplotypes.

Variability of Chloroplast Microsatellite Loci and Population Differentiation in P. brutia, P. nigra, and P. sylvestris

The study of 226 individuals in seven P. brutia populations at variable Pt26081, Pt71936, and Pt15169 loci revealed two, six, and four alleles, respectively (see Fig. 3). The distribution of alleles was uneven. Rare alleles, 1b and 8b, were found at Pt71936 locus only in the Turkish sample. There was no variability at Pt41093 and Pt36480 loci in the studied samples. An average of two alleles per each of the six cpSSR loci (one to four alleles) were found in Black Sea P. brutia. The average intrapopulation diversity was low (H = 0.3657), varying from 0.6090 (Novy Svet) to 0 (Pitsunda). The level of genetic diversity was significantly higher in the Turkish population (H = 0.8857), with an average of 2.33 alleles per locus (from one to six alleles).

In total, ten haplotypes were recorded in seven populations of P. brutia; seven of them were found in Crimean–Caucasian populations and eight in the Turkish sample (see Fig. 1). The allele composition of the haplotypes is shown in Table 2. Three dominant haplotypes (I, II, and IV) are closely related and differ from each other only by one–two mutations. They were found in 92.9% of individuals in the Crimean–Caucasian region . The more divergent haplotype III (four–five mutations from haplotypes I, II, and IV) was found in three populations at a frequency of slightly over 5%; however, its proportion was very high in the Novy Svet (22%). The other haplotypes occurred once or twice. On the whole, the Caucasian populations possessed more haplotypes than the Crimean ones (7 vs. 5) due to rare variants that were identified in Dzhankhot and Arkhipo-Osipovka in the central part of the Caucasian range of P. brutia. Among all populations studied, two Crimean were the most diverged (Fig. 4). Both population samples had four haplotypes with significantly different compositions (see Fig. 1): in addition to haplotype I (57.5%), haplotypes III and II also prevailed in the Novy Svet sample (22.5 and 15%, respectively); in addition to haplotype I (68%), haplotypes IV and II also often occurred in Ayazma (22 and 8%, respectively). Haplotype III was not found in Ayazma and haplotype V was not recorded in Novy Svet.

Table  2. Allele composition of haplotypes of the three variable cpSSR loci in P. brutia populations and the occurrence of haplotypes by region
Full size table
Fig. 4.
figure 4

Ordination of seven populations of P. brutia, constructed on the basis of cpSSR loci, pairwise RST values, and analysis of principal coordinates.

Full size image

The Crimean–Caucasian populations differ slightly from the Turkish sample in haplotype frequencies (see Table 2). The most frequent haplotype I for the Crimean–Caucasian region (75%), which prevails in all samples, was found three times in the Turkish population (20%). The second most frequent haplotype IV (12.3%) was recorded in all populations of the Crimea and Caucasus, except Pitsunda; it was dominant in the Turkish sample (26.7%). One of the haplotypes (II), which was found in three Black Sea populations and had a higher concentration in the Crimea (11.1%), was not observed in the Turkish sample. Three rare haplotypes (VIII, IX, and X) were found only in the Turkish population.

The differentiation of the seven P. brutia populations calculated taking into account differences between haplotypes was RST = 7.47% (see Table 1). The index of RST fixation did not differ significantly from GST (P = 0.6600), which confirms the absence of the phylogeographic structure. The results of the Mantel test are insignificant for P. brutia (P = 0.476). Ordination of populations also does not reveal the geographical structure (see Fig. 4). The subdivision of the six Black Sea populations (without the Turkish one) is even higher (RST = 9.81%).

In the Crimean populations and one Caucasian population of Pinus nigra, three alleles were found at both Pt26081 and Pt71936 loci and six were found at Pt30204 locus; the distribution of alleles was more uniform and unimodal for Pinus nigra than for P. brutia (see Fig. 3). A total of 26 haplotypes were revealed in three samples from 74 trees: the most frequent haplotypes (with a frequency of not less than 6%) were found both in the Crimea and in the Caucasus (in 42% of individuals); 12 haplotypes were recorded in a single sample (four of them in Nikita, three in Ai-Petri, and five in Arkhipo-Osipovka. The P. nigra populations are characterized by an equally high value of genetic diversity (average H = 0.9574); the differentiation of the three populations is RST = 5.77%, with the GST value being insignificant (see Table 1).

Five, three, seven, and six alleles were identified in each of the three Crimean–Caucasian populations of P. sylvestris at Pt15169, Pt26081, Pt30204, and Pt71936 loci, respectively; 87 trees had 39 haplotypes. Unlike P. brutia and P. nigra, it had no dominant haplotypes. Only two most frequent haplotypes were recorded in all populations, and only three haplotypes occurred more than five times. Most of haplotypes (26) were found in one of the populations. The level of cpSSR variability is high for all three populations (average H = 0.958). The differentiation of the studied P. sylvestris populations was GST = 2.16%, with RST being insignificant (see Table 1).

DISCUSSION

High Differentiation without Taxonomic Differences between the Crimean–Caucasian Populations of P. brutia

The study of the cpSSR variability suggests that the Crimean and Caucasian populations of P. brutia are a single group of related populations and have no taxonomic differences from the main range at the species level, which is consistent with the previous data obtained with allozyme markers [17]. Significant differences in the composition of cpSSR alleles and haplotypes were not found between the Turkish population sample and Black Sea populations of P. brutia. The division of the Crimean–Caucasian populations into the species P. stankewiczii and P. pityusa and, accordingly, the hybridization in populations growing from Anapa to Dzhankhot are not confirmed [14].

The high differentiation of P. brutia populations that was determined in the Crimea and Caucasus (RST = 9.81% for the six populations) is explained by the small size of the populations and their long-term isolation from each other under a low gene flow [16]. The haplotypes are randomly distributed between the populations. At the same time, no isolation by distance was detected and populations were not geographically clustered. The differentiation of the P. brutia population, as well as the level of its variation, are obviously determined by the founder effect and genetic drift.

Using the cpSSR markers, we for the first time revealed significant genetic differentiation of P. brutia populations in the Crimea. Two Crimean populations of Ayazma and Novy Svet (so-called Stankewiczii pine and Sudak Pine, which are sometimes classified as P. stankewiczii [14] or subspecies P. brutia ssp. stankewiczii [15]) differ by genetic distances more significantly from each other than from the Caucasian populations classified as P. pityusa according to [14] (see Fig. 4). Despite the significantly higher current abundance and better regeneration of P. brutia in the Ayazma stow, its genetic diversity (H = 0.4922) was lower than that in the Novy Svet population (H = 0.6090). Unlike the allozyme data [18], according to which the difference between the two populations was insignificant and lower than that between the intrapopulation localities, our study showed quite a large difference between the two Crimean populations; at the same time, we also did not reveal any significant difference between the individual localities (these data are not provided here). The marked differences in the composition of haplotypes and the absence of one of the haplotypes (III) in the Ayazma stow, which prevails in Novy Svet, indicate the cessation of gene flow between these populations for a long time, as well as the isolated existence of P. brutia in the two unique Crimean habitats.

Decrease in the Level of Variation in Relict Populations of P. brutia at the Northeastern Limit of Their Range

A sharp decrease in the cpSSR variability was determined in the Black Sea populations of P. brutia (H = 0.3657, an average of two alleles per locus) compared to the population from the main range, in which higher indices of variability were recorded (H = 0.8857, 2.3 alleles per locus), despite the small number of studied individuals; this is consistent with the previous data on the variation of P. brutia in Greece and Turkey [25] (on average, 3.3 alleles per locus were found in this area in the same loci). At the same time, the diversity also decreased in samples of Eldar pine (P. brutia var. eldarica (Medv) Silba) from Azerbaijan and Iran [25] (2.2 alleles per locus). The recorded decrease in the variation of P. brutia var. pityusa, which was not previously revealed by the allozyme data [15], indicates the higher sensitivity of the cpSSR markers than that of nuclear ones to the reduction of populations due to the twofold decrease in the effective population size.

The inconsistency between the abundance and level of variability in a number of P. brutia populations indicates the prevailing influence of historical factors (such as the event of establishment of populations), rather than their current state, on their genetic structure. For instance, the lowest variability was found in a population with a high current abundance (Pitsunda), where the most frequent haplotype I in Crimean-Caucasian populations was fixed. The genetic deficiency of the population in the Pitsunda grove is probably due to the geological history of the seaside spit in the mouth of the Bzyb River, where it grows. It is evident that the spit was formed as a result of the stabilization of the Black Sea level about 5 thousand years ago [43]; consequently, the formation of the river delta and its colonization by P. brutia were historically determined. The relative isolation of the plain population in Pitsunda from P. brutia stands on the mountain slopes led to the founder effect; as a result, the population has only one (most widespread) cpSSR haplotype, despite the high current abundance and good regeneration of P. brutia in the Pitsunda grove.

The significant subdivision of P. brutia populations in the Crimea and Caucasus and low variability of cpSSR are probably due to a strong decrease in their abundance in the past, namely, during the last Pleistocene glaciation or, according to [7], during the marine transgression about 7000 years ago. According to previous allozyme studies [15], the differences between P. brutia in the Crimea and the main range were more significant than the differences between the population in Pitsunda (Caucasus) and the main range. It was assumed that these populations were historically isolated after the reduction and fragmentation of the range of P. brutia in its northeastern part during the Pleistocene, followed by the distribution of the species in the western and southern parts (Asia Minor and the islands of the Aegean and Mediterranean Seas) [15].

The pattern of genetic variation of P. brutia is similar to that for other pine species with limited or fragmentary distribution: P. halepensis Mill., P. pinea L., P. strobiformis Engelm., etc. [24, 27, 44], in which the cpSSR variability is lower and level of population differentiation with respect to the markers under study is much higher than the respective parameters for the widespread species, e.g., P. sylvestris in the main range [31] or Picea obovata [32]. The decrease in the cpSSR variability in the relict populations of P. brutia is also observed compared to the studied pine populations of P. nigra and P. sylvestris from the same areas. It was previously determined [16] that the parameters of allozyme variability were more than two times lower for P. brutia than for P. nigra from the Crimea and P. sylvestris from the main range.

European Black pine, which, similarly to P. brutia, grows in the Crimean–Caucasian region at the extreme northeastern limit of its range, had a high genetic diversity in all populations (average H = 0.9574), which is comparable with the variability of Crimean–Caucasian populations of Scots pine (H = 0.958). In the Crimea, this is due to the abundance of P. nigra on the peninsula, where it prevails among other coniferous species [2, 4] and where the high value of gene flow was previously shown for the populations of this species [16]. The isolation of the Caucasian population, which is locally distributed near the village of Arkhipo-Osipovka [11], did not lead to a decrease in the level of variability of cpSSR loci and its significant differentiation from the Crimean samples, which indicates a steadily high effective population size. Probably the abundance of European Black pine was higher until quite recently than its current abundance in the Caucasus [11]. It should be noted that P. nigra is a more widespread Mediterranean species than P. brutia [3, 33]. According to Naydenov et al. [33], the populations of P. nigra in the Crimea and Caucasus belong to the same unmixed cluster, which combines them with eastern Turkey, while no significant decrease in the variability was observed for European Black pine in the Crimean–Caucasian region.

The average level of variability for the three studied populations of P. sylvestris (H = 0.958) was somewhat lower and their differentiation (GST = 2.16%) was comparable to the previously obtained values for 38 samples of Scots pine from the main part of the range (average H = 0.976, RST = 2.1%) [31]. However, this species, which is more widespread in the Caucasian region than P. brutia and P. nigra (see Fig. 1), requires further research of the variability of cpSSR markers for the entire Crimean–Caucasian range.

Application of Genetic Data to Develop Additional Protection Measures and Prospects for Using the Studied Pine Species in Forest Cultivation

Although the taxonomic (species) differentiation in the Black Sea populations of P. brutia is not confirmed, their significant subdivision due to the decreased gene flow between each other and from the center of the range has been revealed. The historically isolated and genetically differentiated populations in various habitats outside the main range might accumulate adaptations to the conditions of the northeastern limit of their distribution and, consequently, preadaptations to different factors, primarily to low (and periodically negative) winter temperatures. This is probably also true for European black pine. The adaptation of the marginal populations of European black and Calabrian pines to extreme climate conditions for these species requires a careful attention to the protection of their genetic potential. It is important to note that the results should also be taken into account for developing reforestation measures in the south of Russia in conjunction with the problem of global climate change and under the conditions of a possible significant temperature increase in the middle latitudes as early as the 21st century.

Unfortunately, the areas of natural P. brutia and P. nigra stands significantly decrease due to logging, fires, and road construction [11]. Although P. nigra is listed in the Red Data Book of Krasnodar krai, the maintenance of its high genetic diversity requires a total ban on its logging and establishment of specially protected natural areas near the village of Arkhipo-Osipovka, i.e., its habitat in the Caucasus [11]. It should be noted that the natural stands of pines P. nigra and P. sylvestris in the Crimea have the highest level of protection and are protected in several nature reserves [45].

Pinus brutia is listed in the Red Data Book of Russia [5]. In the Caucasus, it is protected in the northern part of its range, namely, in the Bolshoy Utrish Nature Reserve. A large plain area is protected in the southern part of the Caucasian range, namely, in the Pitsunda-Musser Nature Reserve (Abkhazia). At the same time, the significant differentiation of the populations of this species and weak stability of phytocenoses with Calabrian pine [8] also make it necessary to protect some other locations. The results of studying Calabrian pine communities [8] assume the organization of additional special protected natural areas in the central part of the Caucasus range (near Dzhankhot and Arkhipo-Osipovka), which coincides with the area of maximum genetic diversity of P. brutia populations that we determined on the Caucasian coast.

The level of protection of natural populations of P. brutia is obviously also insufficient in the Crimea. The stands of this pine remain part of forestry areas and have the status of forest reserves [45]. The genetic uniqueness of the two Crimean P. brutia populations, which is shown in our study, gives reason to consider them as evolutionarily significant units (ESUs) [28, 46], with the proper level of protection and development of measures for recovering the populations, in particular, taking into account the increase in the recreational press in recent years [47].

Therefore, the revealed low variation and increased differentiation of the Black Sea populations of P. brutia, compared to P. nigra,P. sylvestris, and the P. brutia population from the main part of the range, are presumably due to the small size and long-term isolation of populations from each other, which is, in turn, determined by the narrowness of the coastal zone suitable for P. brutia in this peripheral part of the range. The data that we obtained can be used for developing measures to protect relict populations of P. brutia and P. nigra.