3.1. Genetic Structure of Red Clover Populations
Fingerprinting of Inter Simple Sequence Repeats (ISSR) was used to reveal the DNA diversity of red clover. Overall, 79 fragments were amplified by using five different primers (
Table 2). Those fragments ranged from 450 bp to 2500 bp. ISSR2 primer amplified the most fragments (18), and 77.8% were polymorphic. Meanwhile, primer UBC857 amplified only 13 fragments, of which ten were polymorphic. However, the lowest yield of polymorphic fragments (73.3%) was identified by using the ISSR1 primer. This indicated that all primers were highly specific for polymorphism detection in red clover. The high quality of ISSR primers for red clover polymorphism analysis was also confirmed by polymorphic information content (PIC) values that ranged from 0.4882 (ISSR1) to 0.4922 (ISSR2), with an average of 0.4906. The closer PIC value to 0.5, the more informative and more selective primer is for DNA fingerprinting. In most cases, researchers choose primers with a PIC value close to 0.5 in order to reduce costs and reveal polymorphism with a smaller number of primers [
32]. For instance, Gholami et al. [
33] analysed different orchid species by using ISSR primers who has narrow range of informativeness, and average PIC value range from 0.391 up to 0.488. Therefore, obtained PIC value was in range to use ISSR markers for
Orchidaceae species. However, if the PIC value is below 0.25 it should not be used for DNA fingerprinting, because of its low informativeness [
32].
Analysis of molecular variance (AMOVA) revealed that the differentiation of genetic diversity was higher within populations than among them and reached 83% (
Table 3). This indicated that the majority of the populations were dominated by high genetic diversity within plants and that those populations were heterogeneous. Moreover, the variation in molecular diversity among populations was only 17%. Our results supported previous findings of other researchers. Ulloa et al. (2003) analysed 400 red clover individuals collected in Chile. They found that variation within populations was 80.4%, while between populations, it was 14.1% based on AMOVA results [
34]. A more extensive analysis was performed by Jones et al. (2020). Researchers reported that the variation within red clover populations was 56.3%, while between populations, it was 22.9% [
20]. All these results support the hypothesis that the high distribution of genetic diversity within populations is independent of geographic location [
20,
34,
35]. However, high genetic variation within red clover populations could be justified by the prevalence of seeds [
20,
36]. It is also likely that most of such populations are not isolated by environmental barriers [
37]. Therefore, can crossbreed with different populations (interpopulation crosses), which leads to low genetic diversity among those populations [
22,
37].
Wright (1978) claims, that if the F
ST value of the population is between 0.15 and 0.25, it is highly differentiated [
38]. We have indicated F
ST = 0.182 among populations and the value of the interpopulation indicator PhiPT = 0.173 (
p < 0.001). Therefore, it is likely that the exchange of genetic information among populations is extensive. Meanwhile, gene flow between populations was confirmed by the gene flow index (Nm) value, which indicates the flow of migrants per generation (Nm = 2.2474). In this case, the index is greater than one (Nm > 1); therefore, gene flow is high but enough to negate the effects of genetic drift [
38,
39]. Our findings suggest that the genetic diversity of red clover is declining due to constant gene flow. As a result, some alleles encoding unique traits may already be lost.
The genetic structure of local cultivars and wild populations of red clover has also provided important information about gene flow. The results of Bayesian clustering showed that the populations of
T. pratense best fit two genetic groups when the delta K value was K = 2 (
Figure 2).
Analysis of the population’s genetic structure revealed that most wild populations tended to cluster in the light blue group, while cultivars were clustered in the orange group (
Figure 3). However, several populations have had genetic structures common to both clusters. For instance, individual plants specific to the cultivars were found in wild-type populations. The distribution of such plants was quite diverse among populations. Interestingly, two populations, pop2880 and pop2890, were different from all others. Those two populations were closely related to the cultivars by their genetic structure. Meanwhile, ‘Vytis’ was closely related to the wild type. Such results reveal that the genetic structure of populations and cultivars is not homogeneous. Some cultivars are closely related to populations by their genetic structure, which suggests that cultivars may have spread to wild populations over the years. Gupta et al. (2017) provided similar insights into the genetic structure and diversity of red clover. Researchers have analysed red clover germplasm from the GenBank of the National Temperate Forage Legume Germplasm Unit (USA). They found 91% within and 9% among population genetic variation [
40]. Moreover, the structure of the analysed populations revealed the idea that some populations show considerable admixture in individuals within clusters [
40,
41].
3.2. Genetic Diversity within Populations
The per cent of polymorphic loci (PPL) was determined by Popgene v1.32 software, which allows us to estimate polymorphism within populations [
42]. The results revealed that population pop2887 was divergent from the others by its high polymorphism per cent (PPL = 82.28%), while the lowest polymorphism (PPL = 69.62%) was established in two populations (pop2870 and pop2890) (
Table 4). Meanwhile, ‘Liepsna’ was distinguished by the highest polymorphism (PPL = 83.54%), and ‘Vytis’ stood out by the lowest percentage of polymorphic loci (PPL = 64.56%) among varieties. It is important to note that ‘Vytis’ was bred and listed on the National List of Plant Varieties in 1996, while ‘Liepsna’ was bred in 1957. This assumes that ‘Liepsna’ encountered more random crossbreeds and stronger gene flow interactions over the years.
Nei’s genetic diversity (h) varied from 0.2233 (pop2868) to 0.3008 (pop2887) within populations. The highest Shannon index value (I = 0.4476) was also identified in population pop2887. As a result, this population was dominated by the highest genetic diversity among individual plants. The opposite trend was found in the population pop2868, which had the lowest value of Shannon’s index (I = 0.2233). Therefore, pop2868 was more homogeneous than the rest of the populations, and consequently, the richness of genetic diversity was lower. Osterman et al. (2021) revealed similar results, while analysing Nordic red clover populations. According to Osterman et al. (2021), the percentage of polymorphic loci ranged from 48.2% to 75.9%, but on average, 64.2% [
35].
3.3. Genetic Diversity among Populations
Nei’s genetic distances were calculated to assess the phylogenetic relationship among populations (
Table 5) [
43]. Meanwhile, UPGMA was used to group populations into clusters and to draw a dendrogram. Nei’s genetic distances ranged from 0.041 (pop2887 and pop2879) to 0.141 (pop2875 and pop2870). Populations with the highest Nei’s genetic distance (pop2875 and pop2870) were the least related to all the other populations. Interestingly, pop2870 was very closely related to pop2902 (0.061) and the variety ‘Vytis’ (0.066) according to Nei’s genetic distance. However, pop2902 was closer to the varieties ‘Vytis’ and ‘Sadūnai’ (0.56). This suggests common affinity among varieties and populations (pop2870 and pop2902). However, it is difficult to tell whether those populations and varieties had a common ancestor or whether they had domesticated plants within populations.
UPGMA distinguished two main groups and a few subgroups with short genetic distances from each other. Subgroup IA comprised pop2867 and pop2870 (
Figure 4). Those two populations were assigned to the phenotypic group of semiwilds and did not differ significantly by many of the analysed morphological traits among each other. The following subgroup was composed of two populations (pop2875 and pop2880) and denoted as IC on the UPGMA dendrogram. Neither population pop2875 nor pop2880 produced a second seed yield during the same season and were assigned to the wild phenotypic group.
The largest subgroup was denoted as IB and combined seven populations, as well as four Lithuanian-origin cultivars ‘Liepsna’, ‘Sadūnai’, ‘Kamaniai’, and ‘Vytis’. It is likely that during the breeding process of those Lithuanian-origin cultivars, genotypes of local wild populations could be used. Alternatively, those seven populations have been exposed to gene flow from cultivars.
The cultivar ‘Arimaičiai’ and pop2890 were assigned to subgroup ID. Pop2890 likely has been highly affected by gene flow from cultivars because morphological traits assigned this population to the cultivars’ phenotypic group. In addition, pop2890 stood out by having a higher 1000 seed weight than the other populations [
44]. The wild-type phenotype characterised a separate group of population pop2876. Even though population pop2876 was not unique by its genetic diversity, it was distinguished among other populations by genetic distance. This suggests this population was isolated from random inbreeding with other populations and cultivars. As a result, inbreeding leads to an increase in homozygosity and lethal mutations [
37].
PCA grouped all populations into two distinct groups (
Figure 5). The component axes explained 28.25% of the variation from the data set and showed a partial overlap of these two groups. Resembling results were obtained by Kölliker et al. (2003). The only difference was that the researchers also distinguished a third group of semiwilds [
45]. Although ISSR molecular markers allow us to separate cultivars from populations, they did not provide sufficient information on grouping populations into wild and semiwild populations. This may likely be due to interpopulation crossing, which leads to low genetic distances among cultivars and populations. Either reason may be the relatively small geographical distances between habitats of red clover populations [
20]. Therefore, genotype sequencing for more accurate results should be used to analyse small site populations.
Different opinions prevail among scientists about the relationship of genotypes with their geographical origin. It is likely that the populations analysed in our research were not highly isolated and did not maintain their genetic homogeneity. Therefore, red clover populations and cultivars of local origin were phylogenetically close to each other. Genetic differences mostly depended on the local isolation of populations by natural barriers and the farming type in natural habitats [
46,
47,
48].