Genetic Diversity and Population Structure of Algerian Endemic Plant Species Avena macrostachya Bal. ex Cross. et Durieu

: Avena macrostachya is still a relatively unknown species. Using sequence-related ampliﬁed polymorphism (SRAP) markers, a simple and inexpensive technique, allowed us to conduct experiments on genetic di ﬀ erentiation and to study the population structure of this Algerian endemic oat. The results obtained showed lower than expected genetic diversity within the A . macrostachya species. The cause may be endemism of the species as well as genetic drift possible during collection, and maintenance of the accessions in gene bank and seed reproduction. No clear genetic structure was found in the examined collection, which indicates a close relationship between the populations collected in the Djurdjura National Park in Algeria. Considering the endemism of the species, its breeding potential and the small-scale ex situ collection, careful monitoring of natural sites and repeating of the collection mission are, therefore, absolutely crucial.

The Mediterranean basin is considered a global biodiversity hotspot primarily because of the size and diversity of its flora. About 10% of the world's vascular plants inhabit 1.6% of the world's terrestrial area [13]. In northern Algeria, 22 sites have been identified as Important Plant Areas (IPAs) for endemic plant species. The Atlas Mountains in Algeria include many valuable nature parks and reserves, the Bellezma National Park and the Djurdjura National Park contain the natural sites of A. macrostachya [14].
The Atlas Mountains are located in northern Africa extending over three countries: Morocco, Algeria, and Tunisia, and cover a total area of about 800,000 km 2 . A characteristic feature of the Atlas Mountains are extensive cedar forests with unique species of trees and shrubs, including Cedrus atlantica, Festuca algeriensis Trab., Lonicera kabylica (Batt.) Rehder [15]. Besides, the ecoregion is rich in juniper Juniperus africana (Maire) H. del Villar and evergreen oak forests, Quercus afares Pomel [14]. The region of northern Algeria is rich in over 300 endemic taxa and the total number of taxa of elementary rank is 4000 [16]. A significant factor contributing to endemism is the climate in the Mediterranean basin: Altitude, large thermal amplitudes, west-east rainfall gradient. The geographical configuration, diverse topography and geology of this ecoregion enabled the creation of many microclimates and diverse habitats [17].
Plants of A. macrostachya were found only in two mountain ranges of the Atlas Mountains: The Aures and the Djurdjura Mountains (Table S1). The first region of the occurrence is the whole area of the Aures and Bellezma Mountains and the area of Djebel Chenntgouma, west of Kenchala. Eleven sites where A. macrostachya was found were located at an altitude of 1350-1800 m above sea level, Agronomy 2020, 10,1984 3 of 14 with a strongly moist climate, where grasses and spiny cushion shrubs (e.g., Buplerum spinosum Gouan, Cytisus pungens Spreng.) dominated. The second region was the Djurdjura Mountains, where this oat grew on the meadows of Festuca sp. 15 sites where this species occurred were found at an altitude of 1350-1650 m above sea level. The distribution of A. macrostachya covered a much larger area than in Aureus and Bellezma [18]. In the Djurdjura Mountains, there is a large population of A. macrostachya within the national park. It seems that this species prefers open spaces, such as screes and rocky places, with precipitation >500 mm of annual rainfall and mean minimum temperature of the coldest month of −0.8 • C or lower [1].
Our knowledge of the species, its distribution and variability is very limited. There are 150 records in the Global Biodiversity Information Facility (GBIF) database on the biological resources of A. macrostachya [19]. It includes 82 herbarium specimens and 37 living accessions. A critical analysis of these data narrows the knowledge about the presence of A. macrostachya to around 30 sites (Table S1). The information on A. macrostachya accessions in the other databases is very limited. The FAO database has in its collection only 13 accessions, the majority of which are the accessions preserved in the long-term storage of the Plant Breeding and Acclimatization Institute in Poland (acquisition year 2003). Apart from that, the database contains single accessions of A. macrostachya from Hungary (1995 acquisition year), Canada (1990 acquisition year), and Great Britain (1985 acquisition year) [20].
A. macrostachya is a species attracting special interest of oat breeders as a distinct source of new genes for building resistance to biotic and abiotic stresses [21]. It has been used in some oat breeding programs in UK, USA, and Poland. A. macrostachya accessions survived even severe winters in Canada [2]. The results of Santos et al. [22] confirmed the expected influence of the species variation on improvement of the cultivated oat resistance to low temperatures and the SBMV virus. Transfer of winterhardiness from A. macrostachya is the main objective of the Polish interspecific crossing program aimed at increasing the area of winter oat cultivation [23]. Yu and Herrmann [24] reported introgression of powdery mildew resistance to A. sativa. Loskutov [25] listed other resistances potentially useful for breeders: To crown and stem rust, septoria blotch, helminthosporium blotch, myrothecium blotch, frit fly. Weibull [26] announced resistance to the aphid Rhopalosiphum padi L. Only a small part of the species variation has been applied in oat improvement and the interest of breeders in this source of new genes is increasing. Therefore, better recognition in the level of individuality and population structure of the accessions kept in gene bank collections is important for the future of oat breeding.
The main aim of this work was to determine genetic diversity and population structure of a unique collection of A. macrostachya preserved in the Polish gene bank of the Plant Breeding and Acclimatization Institute (IHAR)-National Research Institute, Radzików.

Plant Material
The plant material used for the study consisted of nine accessions of A. macrostachya preserved by the National Centre for Plant Genetic Resources (NCPGR) in Radzików (Polish gene bank) ( Table 1). Six accessions were acquired by germplasm-collecting expedition carried out by The International Board for Plant Genetic Resources (IBPGR) and Institut National de la Recherche Agronomique de l'Algerie (INRAA) in 1988. One accession (PL 52409) was collected through the NCPGR expedition to Algeria in the year 1989. Another accession (PL 52410) was obtained from the Plant Gene Resources of Canada. The last accession (PL 52412) resulted from open-pollination of PL 52411 and PL 52409. Locations of collection sites of the tested accessions are indicated in Figure 2. with spatial isolation. Morphologically and phenologically all accessions are very uniform. According to Table 1 and geographical coordinates of collection sites distribution area of evaluated accessions covers 257 km 2 (5.55 km SN; 46.25 km WE) ( Figure 2).   The history of maintenance of the accessions is well known, i.e., PL 52403-PL 52409 are first generation of originally collected seed material. PL 52410 is a descendant of seven clumps survived where 30 clones were established. Fifty seeds were received from Plant Gene Resources of Canada (PGRC). Other samples were small and a low number of seedlings were produced in the laboratory. All examined accessions have been preserved in long-term storage and multiplied on multiyear plots with spatial isolation. Morphologically and phenologically all accessions are very uniform.
According to Table 1 and geographical coordinates of collection sites distribution area of evaluated accessions covers 257 km 2 (5.55 km SN; 46.25 km WE) ( Figure 2).

DNA Isolation
Genetic analysis was performed on 64 individuals representing nine accessions. The plant material was collected as young, healthy leaves picked separately from 64 plants. The tissue, immediately after collection, was frozen in liquid nitrogen and further stored in a deep freezer at −80 • C. Before isolation, the tissue was freeze-dried and ground in a bead mill. DNA isolation was carried out using the Genomic Mini Ax Plant kit (A&A Biotechnology, Gdynia, Poland) according to the manufacturer's protocol. A pellet was resuspended in 100 µL of 10 mM Tris-HCl. DNA quality was evaluated using electrophoresis on 1.5% (w/v) agarose gel. Concentration and purity were determined spectrometrically (NanoDrop Spectrophotometer ND-1000, NanoDrop Technologies, Willmington, DA, USA). DNA was diluted in 10 mM Tris-HCl to a working concentration of 25 ng/µL.

SRAP Analysis
A total of 64 combinations of sequence-related amplified polymorphism (SRAP) primer pairs published by Budak et al. [27] were tested in preliminary studies. The eight best combinations were selected for further analysis ( Table 2). PCR amplification was carried out in 25 µL volume of a mixture containing 50 ng DNA, 0.2 u Walk DNA Polymerase (A&A Biotechnology, Gdynia, Poland), 1× reaction buffer (containing 100 mM KCl, 100 mM (NH 4 ) 2 SO 4 , 200 mM Tris-HCl pH 8.5, 20 mM MgSO 4 , 1% Triton X-100), 120 µM of each dNTP and 0.5 µM of each primer (Table 3). Forward primers were labeled with fluorescent dyes on the 5 end (Thermo Fisher, Waltham, MA, USA). The reaction was performed using Verity 96 Thermal Cycler (Applied Biosystems, Foster City, CA, USA under the following temperature profile: 3 min. at 94 • C of initial denaturation, followed by five cycles each one including 1 min at 94 • C, 1 min at 35 • C and 1 min at 72 • C, followed by 35 cycles each one including 1 min at 94 • C, 1 min at 55 • C and 2 min at 72 • C, and the final extension for 10 min at 72 • C. PCR products were separated using capillary sequencer Genetic Analyzer 3130XL (Applied Biosystems, Foster City, CA, USA). The 36 cm capillary array field with NanoPOP7 (Nimagen, Nijmegen, The Netherlands) was used. The length of fragments was assessed against the GeneScan 1200 LIZ Size Standard (Applied Biosystem, Foster City, CA, USA). Each PCR reaction and fragment analysis were tripled. Only repeatable fragments were scored. Table 2. List of sequence-related amplified polymorphism (SRAP) primers used in the study.

Primer
Sequence Dye GAC TGC GTA CGA ATT CTG - Table 3. Primer combinations used in the study.

Data Analysis
Raw data were transformed into a binary matrix where 0 indicated absence and 1 presence of fragment using GeneMapper™ Software v4.1 (Applied Biosystem).
The polymorphism information content (PIC) coefficient for each marker was estimated using the following formula: where i is the ith allele of the jth marker, n is the number of alleles of the jth marker and p is an allele frequency. Unbiased genetic diversity (uH e ) between accessions was estimated as follows: where n is the sample size and pi is the frequency of the ith marker. The significance of differences was tested using ANOVA and post-hoc Tukey's HSD test.
XLSTAT Ecology (Addinsoft, Inc., Brooklyn, NY, USA) software was used to calculate the genetic distance based on the Jaccard coefficient and to perform the Principal Coordinate Analysis (PCoA). GenAlEx 6.501 software was used to perform a hierarchical analysis of molecular variance (AMOVA) [28,29]. STRUCTURE 2.3.4 software was used to determine population structure [30]. The number of group (K) values was set from 1 to 10, with ten independent runs for each K (50,000 burn-ins and 250,000 Markov Chain Monte Carlo generations) and an admixture model with correlated allele frequencies was used. The CLUMPAK software was used to identify the number of real groups in data (K). The optimal value of K was determined based on the a posteriori probability of data for a given K and ∆K using the full search algorithm to obtain the best alignment of the cluster analysis [31,32]. The raw supplementary data are available at https://osf.io/h3erq/.

Marker Informativeness
The results of marker informativeness analysis showed an average number of obtained fragments about 100. The highest number, i.e., 113, was observed for the Me1/Em4 primers combination. The average polymorphism oscillated at the level of almost 70%, the highest, i.e., 81%, was observed for the pair of Me2/Em3 primers. PIC value oscillated on average at the level of 0.227, the highest value was also found for the pair of Me2/Em3 primers (0.270). More details are shown in Table 4.

Genetic Diversity
The value of the genetic diversity coefficient (uH e ) within the studied collection was 0.227. The highest diversity appeared in PL 52405 (0.276) and the lowest in PL 52410 (0.158) ( Figure 3).

Genetic Diversity
The value of the genetic diversity coefficient (uHe) within the studied collection was 0.227. The highest diversity appeared in PL 52405 (0.276) and the lowest in PL 52410 (0.158) (Figure 3).

Population Structure
To examine the relationships of A. macrostachya individuals, PCoA was performed. The first coordinate (Coord. 1), second coordinate (Coord. 2), and third coordinate (Coord. 3) explained 9.12%, 7.29%, and 6.64% of the observed variation, respectively. The gradual overlapping of gene pools of all accessions was clear ( Figure 4). This result is highly consistent with the AMOVA for nine accessions, where 90% of the genetic variation was found among individuals, whereas 10% of the genetic variation was found among accessions (ΦPT = 0.104, p = 0.0001).   Homogeneous groups were labeled with the letters. The significance of differences was tested using ANOVA. (p < 0.05) and post-hoc Tukey's HSD test.

Population Structure
To examine the relationships of A. macrostachya individuals, PCoA was performed. The first coordinate (Coord. 1), second coordinate (Coord. 2), and third coordinate (Coord. 3) explained 9.12%, 7.29%, and 6.64% of the observed variation, respectively. The gradual overlapping of gene pools of all accessions was clear (Figure 4). This result is highly consistent with the AMOVA for nine accessions, where 90% of the genetic variation was found among individuals, whereas 10% of the genetic variation was found among accessions (ΦPT = 0.104, p = 0.0001).
To infer the population structure from genotypic data of the nine accessions, the admixture model in the STRUCTURE software was implemented [30]. The largest ∆K was identified for K = 6, giving the highest a posteriori probability of K [32] (Figure 5). Individuals were assigned into six groups (numbered 1-6) based on a 70% membership threshold. According to this indicator, particular groups from one to nine individuals were assigned (9, 9, 2, 4, 1, and 4 individuals, respectively). The remaining individuals displayed a varying degree and pattern of admixture ( Figure 6). The obtained results indicate that within the accession PL 52403 the 1st group was the most significantly represented. Its total contribution to the accession structure was estimated at 56.5% (Figure 7). In this accession, four of ten studied individuals were classified into the 1st group, one into the 3rd, whilst the others showed admixture. The 1st group also contributed the most to the accession PL 52406 and constituted 58.6% of its entire genetic background. Half of the individuals were classified as members of the 1st group, while the others showed an admixed genetic background. The 2nd group was mainly present in the accessions PL 52404, PL 52410, and PL 52412, and its contribution to the genetic structure was 56.1%, 57.8%, and 50.4%, respectively. Among the 16 individuals studied in a total of these accessions, seven showed membership in the 2nd group. One plant originating from PL 52412 showed a genetic makeup typical for the 3rd group. The remaining individuals were not clearly assigned. The 3rd group appeared as an admixture in all the studied accessions and its contribution ranged from 2.8% (PL 52404) to 28.5% (PL 52410). Out of all 64 tested plants, only two were assigned to this group (from PL 52403 and PL 52412) according to the assumed criteria. The 4th group was the main contributor to the accession PL 52405 (45.4%) and 40% of the individuals have been assigned to it. In addition, one individual was assigned to the 2nd group, and the rest were admixed. The 5th group had the greatest impact on the genetic structure of the accession PL 52408 (39.4%), although all studied plants were considered as having a mixed origin. The last identified in the course of the analysis, the 6th group, prevailed in the genetic structure of the accession PL 52407. Four individuals showed homogeneous background, and the rest were admixed to varying degrees.
To examine the relationships of A. macrostachya individuals, PCoA was performed. The first coordinate (Coord. 1), second coordinate (Coord. 2), and third coordinate (Coord. 3) explained 9.12%, 7.29%, and 6.64% of the observed variation, respectively. The gradual overlapping of gene pools of all accessions was clear (Figure 4). This result is highly consistent with the AMOVA for nine accessions, where 90% of the genetic variation was found among individuals, whereas 10% of the genetic variation was found among accessions (ΦPT = 0.104, p = 0.0001).   To infer the population structure from genotypic data of the nine accessions, the admixture model in the STRUCTURE software was implemented [30]. The largest ΔΚ was identified for K = 6, giving the highest a posteriori probability of K [32] (Figure 5). Individuals were assigned into six groups (numbered 1-6) based on a 70% membership threshold. According to this indicator, particular groups from one to nine individuals were assigned (9, 9, 2, 4, 1, and 4 individuals, respectively). The remaining individuals displayed a varying degree and pattern of admixture ( Figure 6). The obtained results indicate that within the accession PL 52403 the 1st group was the most significantly represented. Its total contribution to the accession structure was estimated at 56.5% (Figure 7). In this accession, four of ten studied individuals were classified into the 1st group, one into the 3rd, whilst the others showed admixture. The 1st group also contributed the most to the accession PL 52406 and constituted 58.6% of its entire genetic background. Half of the individuals were classified as members of the 1st group, while the others showed an admixed genetic background. The 2nd group was mainly present in the accessions PL 52404, PL 52410, and PL 52412, and its contribution to the genetic structure was 56.1%, 57.8%, and 50.4%, respectively. Among the 16 individuals studied in a total of these accessions, seven showed membership in the 2nd group. One plant originating from PL 52412 showed a genetic makeup typical for the 3rd group. The remaining individuals were not clearly assigned. The 3rd group appeared as an admixture in all the studied accessions and its contribution ranged from 2.8% (PL 52404) to 28.5% (PL 52410). Out of all 64 tested plants, only two were assigned to this group (from PL 52403 and PL 52412) according to the assumed criteria. The 4th group was the main contributor to the accession PL 52405 (45.4%) and 40% of the individuals have been assigned to it. In addition, one individual was assigned to the 2nd group, and the rest were admixed. The 5th group had the greatest impact on the genetic structure of the accession PL 52408 (39.4%), although all studied plants were considered as having a mixed origin. The last identified in the course of the analysis, the 6th group, prevailed in the genetic structure of the accession PL 52407. Four individuals showed homogeneous background, and the rest were admixed to varying degrees.

Marker Effectivness
Here, the study of genetic diversity was performed using the sequence-related amplified polymorphism (SRAP) technique, which was developed by Li and Quiros in 2001 [20]. SRAP uses a combination of 17 or 18 nucleotides "forward" primers from GC-rich exons and "reverse" primers from AT-rich promoters or introns and spacers. SRAP technique is based on the amplification of open reading frames (ORF), which may strengthen the relationship between DNA polymorphisms and morphological characteristics [21]. The reproducibility of these markers is much better than that of the RAPD technique, and an additional advantage of the SRAP technique is the possibility to use fluorescent labeled primers and combine them with unlabeled primers to separate products in capillary sequencers [22]. Another advantage of SRAP is a small amount of starting DNA necessary for analysis (approx. 50 ng). The system does not require knowledge of the sequence or reference genome information. The SRAP markers were used previously for assessing genetic relationships

Marker Effectivness
Here, the study of genetic diversity was performed using the sequence-related amplified polymorphism (SRAP) technique, which was developed by Li and Quiros in 2001 [20]. SRAP uses a combination of 17 or 18 nucleotides "forward" primers from GC-rich exons and "reverse" primers from AT-rich promoters or introns and spacers. SRAP technique is based on the amplification of open reading frames (ORF), which may strengthen the relationship between DNA polymorphisms and morphological characteristics [21]. The reproducibility of these markers is much better than that of the RAPD technique, and an additional advantage of the SRAP technique is the possibility to use fluorescent labeled primers and combine them with unlabeled primers to separate products in capillary sequencers [22]. Another advantage of SRAP is a small amount of starting DNA necessary for analysis (approx. 50 ng). The system does not require knowledge of the sequence or reference genome information. The SRAP markers were used previously for assessing genetic relationships

Marker Effectivness
Here, the study of genetic diversity was performed using the sequence-related amplified polymorphism (SRAP) technique, which was developed by Li and Quiros in 2001 [20]. SRAP uses a combination of 17 or 18 nucleotides "forward" primers from GC-rich exons and "reverse" primers from AT-rich promoters or introns and spacers. SRAP technique is based on the amplification of open reading frames (ORF), which may strengthen the relationship between DNA polymorphisms and morphological characteristics [21]. The reproducibility of these markers is much better than that of the RAPD technique, and an additional advantage of the SRAP technique is the possibility to use fluorescent labeled primers and combine them with unlabeled primers to separate products in capillary sequencers [22]. Another advantage of SRAP is a small amount of starting DNA necessary for analysis (approx. 50 ng). The system does not require knowledge of the sequence or reference genome information. The SRAP markers were used previously for assessing genetic relationships and diversity within and between Saccharum genera [23], Brassica napus L. [24], Cynodon dactylon L. [25], Buchloë dactyloides Nutt [26], Cucurbita moschata Duchesne [27], Pinus kesiya Royle ex Gordon var. langbianensis (A. Chev.) Gaussen [28]. SRAP markers amplify coding regions, which may result in their lower variability [34]. Research on the genetic diversity of the Egyptian pomegranate (Punica granatum L.) indicated SRAP markers had higher sensitivity, PIC and discrimination capacity comparing to ISSR and ISTR markers [35]. The effectiveness of SRAP markers for the genetic variation studies in the genus Avena was confirmed during the evaluation of 64 Avena strigosa Schreb. accessions [36]. The pattern of variation level within the groups was the same regardless of the method of analysis, i.e., morphological traits, isoenzymes, or SRAP markers. Based on the SRAP analysis results, it was also possible to identify the loss of genetic integrity of some accessions due to genetic drift [36]. In the presented study, eight combinations of SRAP primers pairs allowed the identification of 781 fragments of which 69% were polymorphic. Both of these values were higher than previously obtained for A. strigosa [36]. A higher level of polymorphism detected in the study of A. macrostachya comparing to A. strigosa may resulted both from a different flowering biology (crossvs. self-pollination) and a different approach to the collection and analysis of samples (individuals vs. bulk samples). The mean value of PIC (0.227) in comparison to the analysis of A. strigosa (0.33) was lower [36]. This is a direct result of the high proportion of rare fragments with low frequency. In the studied accessions of A. macrostachya, their average contribution was 27.8%. According to Anderson et al. (1992) formula for dominant markers PIC range 0-0.5 [37]. Therefore, its value for the SRAP marker will be the average of the PIC values obtained for all fragments generated by a given pair of primers. Therefore, when a pair of primers detects a large number of unique alleles for which PIC is very low, the average value will be significantly decreased by them.

Genetic Diversity
The overall genetic diversity in the examined A. macrostachya collection at the level of 0.23 does not differ significantly from the values observed in previous studies of other Avena species. In the study of the eight accessions of Avena strigosa Schreb., the average value of the uHe coefficient was 0.21, whereas the genetic diversity of Polish landraces of Avena sativa L. was determined as 0.23 [38,39]. Significantly lower values of the differentiation were observed for common oat commercial cultivars, i.e., 0.07 [40]. However, the three oat species differ in terms of their biological status, ploidy level, and also breeding system and each of these factors substantially affects genetic diversity [41]. Generally, genetic diversity increases with ploidy [42]. This phenomenon was observed in the family Poaceae for Cynodon and B. dactyloides [43][44][45]. The three previously mentioned oat species represent three subsequent levels of ploidy, i.e., A. strigosa is a diploid species (2n = 14), A. macrostachya is a tetraploid (2n = 28) and A. sativa is a hexaploidy (2n = 42). The biological status, indicating whether the accessions under investigation represent wild species, landraces, or cultivars, is directly related to the expected degree of their genetic diversity. Along with the domestication of the species, a decrease in their genetic diversity is observed [46]. Accessions of A. macrostachya studied here reflect populations of the wild species, while both A. strigosa and A. sativa are domesticated species. Crops show a decreased diversity level caused by a serious population size reduction resulting from sub-sampling of wild ancestral species followed by targeted selection for the desired phenotypic characteristics during the breeding. Plant breeding systems have been identified as a major factor influencing genetic diversity. Cross-pollinated species are expected to have a higher level of genetic diversity compared to self-pollinating ones. In general, outcrossing reduces the relationship between uniting gametes, thus increasing the number of effective alleles and blurring population subdivision [47]. A. macrostachya is the only outcrossing species of the genus Avena. [40,46]. Considering all the above-mentioned facts, it was expected that the genetic diversity of the examined accessions of A. macrostachya would be higher than that observed in cultivated oats. Relatively low genetic differentiation may result both from the specificity of the species and the fact that the investigated material was acquired from a gene bank. Thus, the loss of diversity may have resulted from a genetic drift that may have been caused by insufficient seed collection during the gene bank expedition, or the loss of alleles during the reproduction and regeneration of the gene bank accessions. Genetic diversity could also be lost due to the deterioration of seed viability during ex situ storage. In the absence of comparative material from natural sites, it is difficult to determine whether, and if so, to what extent, genetic diversity has been lost. According to Baum and Rajhathy [2], in genus Avena, a number of species seem to have a relict endemic distribution limited to isolated places in high elevations or on islands. As mentioned in the introduction A. macrostachya is an endemic species, occurring only in the Atlas Mountains. Populations of endemic species are more vulnerable to genetic drift and reduction of diversity [47]. Since the A. macrostachya sites occur in the mountains at an altitude exceeding 1300 m above sea level, natural selection related to the severe climate should also be considered as a diversity reducing factor. Furthermore, it seems to have a disjunct distribution (Aures and Djurdjura), which can be looked at as an intermediate stage of relict endemism in a process towards extinction. In the future, some of these small populations may become extinct, resulting in relict endemism in advanced stages of this reduction process. Thus, A. macrostachya appears to be one of the morphologically and geographically most isolated species in Avena and it probably originated from a very early branching off of the primitive Avena stock [2]. However, field studies in the Djurdjura showed that the partially isolated A. macrostachya populations were relatively numerous, so the low degree of variation was not caused by the bottleneck effect [1].

Population Structure
Knowledge about the population structure and diversity distribution, i.e., whether it is located within or between populations of the tested species, is important for preserving its genetic diversity [48]. In the study, AMOVA revealed no significant differentiation between the accessions (ΦPT = 0.094). Uneven distribution of genetic variation towards the intra-population one is typical for wind-pollination outcrossing species because this type of breeding system and pollination mechanisms promotes heterozygosity of the individuals [49][50][51][52][53]. The analysis of the genetic structure showed that there is no clear distinction between the accessions representing natural populations. The gene pools permeated each other, indicating some gene flow between the populations. The gene flow between populations is most probably caused by the pollination mechanism. Pollen transported with the wind can reach long distances, eliminating the subdivisions [47]. As is generally known, even a small gene flow across distant populations leads to atrophy of their differentiation [54,55]. Another possible factor that increases gene flow is the transmission of grains by animals. Cattle pasturing was allowed in the Djurdjura National Park so that the seeds could be transported on cattle hair [1]. Descriptions of sites in the Djurdjura National Park made by Guarino et al. in 1988 [1] indicated the continuity of A. macrostachya in this region. The lack of a definite fragmentation of sites in a short time before seed collection to the Polish Gene Bank seems to have the greatest impact on the absence of a distinctive population structure.

Conclusions
To summarize, the research presented in this paper for the first time describes genetic diversity within the A. macrostachya species. Relatively low diversity value may be a consequence of both the endemic character of the species and the genetic drift during collection and preservation in the gene bank. No clear genetic structure has been found within the examined collection, which indicates that the gene flow between the populations in the Djurdjura National Park has been continuously maintained in the late 1980s.
The importance of this research is underlined by the applicability of A. macrostachya in common oat breeding and the uniqueness of the studied collection. It is extremely important to re-establish the expedition in order to identify the changes that have occurred at natural sites. The molecular study of A. macrostachya confirmed the genetic uniformity of the species as already evidenced by morphological uniformity Genetic analysis of the currently occurring specimens is necessary to determine the effectiveness of in situ an ex situ conservation of this species. Considering the endemism of the species, its breeding potential, and the small-scale ex situ collection, careful monitoring of natural sites and repeating of the collection mission is therefore absolutely crucial.