Genetic Differentiation of Red Clover (Trifolium pratense L.) Cultivars and Their Wild Relatives ^†

Abstract

Biodiversity of red clover is changing. It is become more difficult to find new breeding material. Moreover, populations with predominant morphological features of cultivars are more frequently found in wild habitats. Therefore, it is important to identify genetic structure and diversity within and among wild populations and cultivars worldwide. During this research, seed samples from wild red clover (Trifolium pratense L.) habitats were collected in different Lithuanian sites in nemoral environmental zone. Subsequently, samples were germinated in greenhouse and randomly taken for genetic analyses. ISSR fingerprints of 339 genotypes from 13 populations and 5 varieties were analysed. Analysis of molecular variance (AMOVA) has shown high genetic variation within populations (83%). The highest polymorphism was identified in population pop2887 (PPL = 82.28%). Interestingly the polymorphism of varieties was high and ranged from 64.56 (‘Vytis’) to 83.54 (‘Liepsna’). Meanwhile, variance among populations was low—17%. Fixation index FST = 0.182 revealed high differentiation of the populations while gene flow was high with an exchange of 2.2474 migrants per generation. This revealed a high genetic diversity within populations and high gene flow between wild populations and cultivars in Lithuania.

1. Introduction

Red clover is one of the most widely grown species of the Fabaceae family for animal feed in nemoral environmental zone [1,2,3]. The name of Trifolium, translated from Latin means: three from “tres” and leaf from “folium”. This indicates the main distinguishing feature of this genus—trifoliate foliage; however, four-leaf- or even five-leaf red clovers may be found [4,5]. It has been reported that four-leaf clover plants occur at a frequency of 1:6000, while five-leaf clovers could be found in 1 of 10,000 wild populations [5]. Another original feature of Trifolium pratense L. is its V-shape white mark on the leaves [5,6]. Those triangular markings are distinguishing characteristics that allow us to identify species within the genus Trifolium [5,7,8]. Nevertheless, more phenotypic characteristics assist in identifying species from the same genus. However, excluding cultivars among their wild relatives is more difficult, especially at the full seed maturity stage or when plants are defoliated. Therefore, genetic analyses are indispensable for better taxonomy identification.

The abundance of red clover varieties bred in recent decades reveals high demand for this species. During modern breeding they became more productive and more resistant to pests and abiotic stress. However, new varieties remain more or less the same feature—to produce high-quality fodder yield. In addition, red clover is appreciated as an excellent intercrop and crop rotation species for sustainable farming or “greening” programs implementation [9,10]. Moreover, symbiosis with root-nodule bacteria allows red clover to capture atmospheric nitrogen and accumulate it in the soil, which makes this species so tempting to farmers [4,10,11].

Red clover is a diploid species by its nature, and cells of autochthonous red clover have a diploid chromosome set (2n = 2x = 14), while the breeders present cultivars that are modified to have a tetraploid set of chromosomes (2n = 4x = 28) [4,12,13]. Different inhibitors (colchicine, amiprophos-methyl, trifluralin, or oryzalin) may induce polyploidisation [14,15]. However, the methods of breeding are gradually changing. The process has become more diversified compared to the last century. Recently, red clover has been used in the food industry as a bioactive supplement. Moreover, this species is cherished by beekeepers and landscapers [16,17]. This shows a wide application of Trifolium pratense L.

New breeding directions demand deeper knowledge of the genetic structure of the species. Moreover, plant genetic resources are an important element of biodiversity and national assets that must be preserved. It is crucial to know where genetic resources that meet breeding aims can be found. The process of losses and fragmentation of natural habitats has been particularly intensified over the last century. Researchers have noticed that biodiversity is declining in different species, including red clover [18,19]. The gene flow of domesticated red clover likely causes some changes in the genetic diversity of wild populations. In addition, it leads to changes in allele frequencies and a decrease in the gene pool size [20]. Moreover, crop wild relatives (CRWs) are suppressed by inbreeding with cultivars [21,22]. As a result, productivity and resistance to diseases decrease in relation to plants’ genetic diversity.

It is important to identify whether there are autochthonous red clover populations in wild grasslands in Lithuania. Moreover, it is crucial to identify the types of populations that currently predominate in natural habitats and the structure of those populations. This knowledge would be useful for understanding gene flow dynamics between cultivars and their wild relatives. Moreover, it would be useful to identify a new gene donor for breeding because new genetic material is the basis for breeding programs.

This study aimed to determine the genetic structure of wild red clover populations and local cultivars in Lithuania, which is in the nemoral environmental zone. The genotypic diversity of red clover populations collected across Lithuania was compared with locally bred varieties. Furthermore, the links between genetic diversity and gene flow were evaluated.

2. Materials and Methods

At the beginning of the research, seeds from wild red clover populations were collected across Lithuania in 2016. Those seeds were collected from 32 wild and semiwild populations and two from Plant Gene Bank (Figure 1). Five red clover varieties bred in Lithuania were also included in the project.

In the second step, seeds of red clover populations and varieties germinated in the greenhouse on 18 April 2017. Each population was composed of at least 45 individual plants. Samples for genetic analysis were taken randomly from 24 plants in each population. Each sample for at least 100 mg was taken from young and fresh red clover leaves. The samples were frozen in a deep freezer at −80 °C and stored till genetic analysis in the laboratory was started. Finally, red clover plants of each population and variety were transplanted into the field trial ex situ on 10 July 2017.

Based on phenotypic differences, all the populations were divided into three groups: cultivars, wilds, and semiwilds (Figure A1) [23].

Populations for further genetic analysis were chosen randomly from each of those three groups. The ISSR technique analysed a total of 13 populations and 5 Lithuanian-bred varieties (

Table A1. Origin, sampling size, population code and location of Trifolium pratense L. populations and cultivars.

Pop. Code	Pop. Size., No	Origin	Latitude (N)	Longitude (E)
‘Arimaičiai’	23	Variety (LT)	–	–
‘Kamaniai’	16	Variety (LT)	–	–
‘Liepsna’	18	Variety (LT)	–	–
‘Sadūnai’	20	Variety (LT)	–	–
‘Vytis’	15	Variety (LT)	–	–
pop2867	18	Raseiniai distr., Maslauskiškiai.	55.36811°	23.32267°
pop2868	24	Šilalė distr., Karūziškės II, Medvėgalis mound.	55.62802°	22.39114°
pop2869	22	Šilalė distr., Paršežeris.	55.64130°	22.30454°
pop2870	20	Telšiai distr., Mažieji Burbiškiai, Moteraitis mound.	55.76428°	22.49596°
pop2871	18	Jurbarkas distr., Greičiai.	55.05878°	22.64937°
pop2875	21	Telšiai distr., Šatrija mound.	55.87134°	22.55596°
pop2876	18	Telšiai distr., Badmakiai.	55.96160°	22.52041°
pop2877	15	Kėdainiai distr., Čystapolis.	55.47403°	23.66135°
pop2879	16	,Panevėžya distr., Butrimoniai.	55.49606°	24.12171°
pop2880	22	Panevėžiys distr., Vadatkėliai.	55.64402°	24.12658°
pop2887	17	Population from plant gene bank.	–	–
pop2890	14	Population from plant gene bank.	–	–
pop2902	22	Lazdijai distr., Veisiejai region park.	54.12698°	23.66182°

). Genomic DNA was extracted from 100 mg of thawed leaves based on the CTAB method [24].

DNA concentration and quality were evaluated using NanoDrop 2000 spectrophotometer (Thermo Scientific, Waltham, MA, USA). Only samples with DNA concentrations of at least 50 ng μL⁻¹ and quality between 1.7 and 1.9 spectrophotometer laser wavelength ratios (A260/A280) were used for further research.

Polymerase chain reactions (PCR) were repeated at least twice. A reaction mixture of 10 μL was used to perform ISSR-PCR for each genotype. Only vivid and well-reproducing DNA fragments were evaluated.

ISSR-PCR was performed using Mastercycler^® and Mastercycler Gradient^® (Eppendorf, Germany) thermocyclers. The most efficient primers with clarity and repeatability and an abundance of fragments were selected from 30 primers and used for further analyses according to previous research [25,26]. All populations were analysed by using five primers (

Table 1. Sequences of ISSR primers.

Primer	Sequence 5′→ 3′	Annealing °C
ISSR1	AGAGAGAGAGAGAGAGAGAGC	52
ISSR2	AGAGAGAGAGAGAGAGAGAGT	50
UBC827	ACACACACACACACACG	53
UBC857	ACACACACACACACACCG	55
155H	CACACACACACACAGA	49

).

Amplification of ISSR fragments: initial DNA denaturation at 95 °C for 3 min followed by 30 cycles: DNA denaturation at 95 °C for 30 s, next step annealing of primers for 30 s at 49–55 °C. In the penultimate step, DNA was synthesised at 72 °C for 60 s, and the final synthesis was performed at 72 °C for 5 min. ISSR-PCR products were run on 1.5% agarose gel. Finally, gels were analysed under UV light using the MiniBIS pro (DNA Bio-Imaging Systems) documentation system.

Statistical Analysis

Several specific programs were used for data processing and statistical analyses. The unweighted pair group method with arithmetic mean (UPGMA) was used to determine phylogenetic affinity. Population polymorphism, F-statistics, Nei’s genetic distances, and molecular genetic diversity analysis (AMOVA) were performed using the GenAlEx 6.503 program. Principal component analysis (PCA) was performed using SAS 9.4 software [27]. Nei’s genetic diversity (h), Shannon diversity index (I), and percentage of polymorphic fragments (PPL) were calculated using the Popgene v1.32 computer program.

The genetic structure of the populations was analysed by Structure v2.3.4 software using the Markov Chain Monte Carlo (MCMC) algorithm. The value of the selected number of groups (K) was not lower than the number of analysed populations (K = 1–18). The procedure used the Admixture model, and the test repetition period was 100,000 and 100,000 MCMC iterations with three independent runs. The structure Harvester program determined the optimal K value based on the maximum ΔK value. The genetic structure model was developed by the Clumpak program based on the results of Bayesian analysis [28].

The fixation index F_ST was calculated to assess the genetic differentiation of populations:

F_ST = (H_T − H_S)/H_T,

where H_T is expected heterozygosity and H_S—observed heterozygosity [29].

The value of polymorphism information content (PIC) was calculated according to the formula:

PIC = 1 − Σ p_i² − Σ Σ p_i² p_j²,

where p_i and p_j are the population frequency of the i-th and j-th alleles [30,31].

3. Results and Discussion

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. The amplification results of ISSR primers.

Primer	Fragment Size (bp)	No. of Loci	No. of Polymorphic Loci	Percentage Polymorphic Loci (PPL) %	Polymorphic Information Content (PIC)
ISSR1	500–2500	15	11	73.3	0.4882
ISSR2	500–2500	18	14	77.8	0.4922
UBC827	600–2500	17	13	76.5	0.4911
UBC857	600–2400	13	10	76.9	0.4915
155H	450–2300	16	12	75.0	0.4898
Total/avg.	– –	79/15.8	60/12	–/75.9	–/0.4906

). 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. Analysis of molecular variance (AMOVA) in red clover populations.

	df	SS	MS	Est. Var.	Variation %	PhiPT **	Nm
Among Pops	17	948.549	55.797	2.364	17%	0.173	2.2474
Within Pops	321	3637.776	11.333	11.333	83%
Total	338	4586.324		13.697	100%

** p < 0.001 df—degrees of freedom; SS—sum of squares; MS—mean square; Est. Var.—estimated variation; PhiPT—Phi statistics, Nm—number of migrants.

). 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. Genetic diversity of red clover populations based on ISSR analysis.

Population/Variety	N	Na	Ne	PL	PPL %	h	I
2867	18	1.608	1.426	59	74.68	0.2511	0.3781
2868	24	1.506	1.365	57	72.15	0.2233	0.3431
2869	22	1.595	1.422	59	74.68	0.2484	0.3749
2870	20	1.506	1.440	55	69.62	0.2566	0.3813
2871	18	1.582	1.450	60	75.95	0.2635	0.3940
2875	21	1.570	1.474	60	75.95	0.2728	0.4047
2876	18	1.608	1.418	59	74.68	0.2499	0.3778
2877	15	1.532	1.469	59	74.68	0.2736	0.4067
2879	16	1.557	1.424	59	74.68	0.2511	0.3794
2880	22	1.544	1.450	59	74.68	0.2627	0.3926
2887	17	1.709	1.518	65	82.28	0.3008	0.4476
2890	14	1.506	1.384	55	69.62	0.2294	0.3471
2902	22	1.671	1.491	64	81.01	0.2854	0.4258
‘Arimaičiai’	23	1.595	1.458	60	75.95	0.2666	0.3980
‘Kamaniai’	16	1.658	1.471	63	79.75	0.2767	0.4145
‘Sadūnai’	20	1.494	1.446	57	72.15	0.2588	0.3853
‘Vytis’	15	1.430	1.408	51	64.56	0.2360	0.3509
‘Liepsna’	18	1.722	1.512	66	83.54	0.2988	0.4465

N—population size, Na—the average number of alleles, Ne—effective number of alleles, PL—number of polymorphic loci, PPL—the percentage of polymorphic loci, h—Nei (1973) genetic diversity, I—Shannon index.

). 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. Pairwise population matrix of Nei’s genetic distance of red clover.

	2867	2868	2869	2870	2871	2875	2876	2877	2879	2880	2887	2890	2902	ARI	KAM	SAD	VYT	LIE
2867		0.071	0.102	0.065	0.069	0.130	0.106	0.080	0.086	0.128	0.078	0.113	0.083	0.124	0.106	0.084	0.086	0.087	2867
2868			0.070	0.076	0.062	0.092	0.109	0.064	0.057	0.103	0.058	0.090	0.082	0.087	0.085	0.075	0.102	0.090	2868
2869				0.092	0.095	0.079	0.086	0.072	0.096	0.105	0.091	0.104	0.063	0.096	0.104	0.088	0.077	0.089	2869
2870					0.066	0.141	0.115	0.080	0.079	0.119	0.087	0.133	0.061	0.095	0.095	0.091	0.066	0.083	2870
2871						0.088	0.125	0.069	0.050	0.086	0.068	0.136	0.073	0.125	0.091	0.087	0.080	0.088	2871
2875							0.109	0.081	0.068	0.076	0.075	0.108	0.080	0.116	0.083	0.070	0.100	0.093	2875
2876								0.111	0.112	0.129	0.108	0.130	0.107	0.124	0.120	0.112	0.114	0.093	2876
2877									0.059	0.095	0.050	0.058	0.068	0.083	0.066	0.074	0.070	0.073	2877
2879										0.072	0.041	0.091	0.065	0.092	0.088	0.068	0.063	0.065	2879
2880											0.093	0.101	0.060	0.103	0.071	0.078	0.098	0.083	2880
2887												0.084	0.073	0.078	0.085	0.061	0.065	0.059	2887
2890													0.098	0.101	0.111	0.098	0.101	0.103	2890
2902														0.081	0.074	0.056	0.056	0.068	2902
ARI															0.102	0.104	0.088	0.073	ARI
KAM																0.058	0.090	0.073	KAM
SAD																	0.075	0.056	SAD
VYT																		0.072	VYT
LIE																			LIE

ARI—‘Arimaičiai’, KAM—‘Kamaniai’, SAD—‘Sadūnai’, VYT—‘Vytis’, LIE—‘Liepsna’.

) [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].

4. Conclusions

High genetic diversity was found within red clover populations and local varieties. Meanwhile, genetic distances among populations were low. A high affinity of local red clover varieties with wild populations was also found, especially with the oldest variety, ‘Liepsna’. This shows that there is an interpopulation crossbreeding between populations and cultivars. Moreover, the exchange of genetic information among red clover populations and cultivars is extensive in small site areas like Lithuania. As a result, high differentiation and high gene flow are dominant among populations.