Molecular diversity and genetic relationships among Geranium pusillum and G. pyrenaicum with inter simple sequence repeat (ISSR) regions

ABSTRACT

Successful management and preservation of natural populations depend on accurate assessment of genetic diversity. Knowing the genetic diversity within a population is important for choosing the conservation strategies for the species. Geranium pusillum L. is an annual or biennial herb mainly native to Europe, northern Africa, and western Asia, and G. pyrenaicum Burm is a frost hardy perennial native to areas of France and Spain to the Caucasus. In general, taxonomic and biosystematic studies of Geranium are not known in Iran; moreover, in a few cases, inter-specific hybrids and intermediate forms are recognized. A detailed morphological, micro-morphological and molecular (ISSR) study of Geranium is done here with the following objectives: (1) to delimitate the species; (2) to carry out a population genetic study and produce information on genetic structure and genetic variability within each population in G. pyrenaicum and G. pusillum. The present study revealed that a combination of morphological and micro-morphological (seed and pollen) data can separate the species. In the present study, 84 randomly collected plants from 12 geographical populations of two Geranium species were considered. A consensus tree based on morphological and genetic data separated some of these populations from the others, suggesting the existence of ecotypes in the studied taxa.

KEYWORDS:

• Geranium
• ISSR
• morphology
• species delimitation

Introduction

Understanding the genetic variation within populations is important in choosing conservation strategies for plant species, as the preservation of genetic diversity of the populations is a fundamental goal of conservation biology (Mulligan and Findlay 1970). The maintenance of genetic diversity has a considerable significance for the survival of endangered plant species (Mulligan 1974). The ability of a species to adapt to environmental changes depends greatly on the genetic diversity in its gene pool (Nebauer et al. 1999). Geranium L. is a cosmopolitan genus, with about 320 species being recognized in three subgenera and 18 sections (Aedo 2017). A brief history of the generic delimitation and infra-generic classification, and a complete description of the genus are provided by Aedo (1996). An identification key for subgenera and sections are provided as well (Aedo et al. 1998a, 1998b). Geranium subg. Robertium (Geraniaceae) comprises eight sections, of which Batrachioidea W.D.J. Kotch consists of four species centered in Eurasia, between the Mediterranean region and the Himalaya Mountains. Most Geranium species are herbaceous perennial plants with horizontal rhizomes, but some species are annual. Three species, Geranium pusillum L., G. molle L., and G. pyrenaicum Burm. f., show some degree of morphological overlaps and potential inter-specific hybridization; therefore, their taxonomic delimitation becomes somewhat problematic, particularly in areas that overlap in the vicinity. Two of these species, i.e. Geranium pusillum and G. pyrenaicum, occur in Iran.

Inter-specific hybridization is known to occur in the genus Geranium; for example, hybrids have been described as G. × oenense (G. molle × G. pusillum); G. × luganense (G. molle × G. pyrenaicum); and G. × hybridum (G. pusillum× G. pyrenaicum) (Van Loon 1984c).

Plant species delimitation and infra-specific genetic diversity are two important areas of investigation in phylogenetic systematics, evolution, biogeography and biodiversity studies. Data obtained can help to understand the patterns and mechanisms of speciation and hybridization (Esfandani Bozchaloyi et al. 2017a, 2017b). They can reveal the pattern of gene flow between closely related phylogenetic species versus isolation by distance and identify the evolutionary process by which new biological species arise (Freeland et al. 2011). Species delimitation is a difficult task, particularly in species with cross-pollination breeding systems that tend to form frequent inter-specific hybrids (Esfandani Bozchaloyi et al. 2017b, 2018a, 2018b).

Field species recognition and sampling are generally based on both discrete qualitative morphological characters as well as quantitative morphological characters that show no overlap with other species. In many cases, this approach fails to discriminate species and masks the presence of cryptic species, or discriminates different species while in reality there is only one (Wiens 2007). Therefore, in recent years detailed population based study and sampling have been preferred for species delimitation, and a combination of different approaches and molecular data have been used to delimit these taxonomic identities (Esfandani Bozchaloyi et al. 2017b, 2017c, 2017d). Species delimitation in Geranium Sect. Batrachioidea using a combination of morphological, and molecular markers like ISSRs (inter-simple sequence repeats), ITS (internal transcribed spacer DNA of ribosomal DNA) and CP-DNA has been carried out in Iran (Esfandani Bozchaloyi et al. 2017b).

In the present study a combination of micromorphology (seed and pollen) and morphology is used to carry out species delimitation. For this purpose, we collected plants of Geranium pusillum and G. pyrenaicum from the areas they grow and the areas of overlap and delimited these two species. Studying intra-specific variability in these two species is another aim of the present study; therefore, we undertook population genetic investigation of the mentioned species.

Intra-specific genetic diversity has a considerable significance for the survival of plant species (Mulligan 1974), and the ability of a species to adapt to environmental changes (Nebauer et al. 1999). For both species delimitation and population genetic studies, a combination of different molecular markers have been used (see for example Esfandani Bozchaloyi et al. 2017c, 2017d).

Materials and methods

Plant materials

A total of 84 individuals were sampled representing 12 natural populations in East Azerbaijan and Mazandaran Provinces of Iran (Table 1). Forty-nine plant samples were collected from seven geographical populations of G. pyrenaicum and 35 plant samples were collected from five geographical populations of G. pusillum. Different references were used for the correct identification of species (G. pyrenaicum Burm and G. pusillum L.), (Davis 1965; Schönbeck-Temesy 1970; Zohary 1972; Aedo et al. 1998a; Janighorban 2005). Details of sampling sites are mentioned in Table 1 and Figure 1. Voucher specimens are deposited in the Herbarium of Shahid Beheshti University (HSBU).

Table 1. Location and herbarium accession numbers of the studied populations of G. pyrenaicum Burm and G. pusillum L. collected by Esfandani in Iran.

Species	Population	Locality	Latitude	Longitude	Altitude (m)	Voucher no.
G. pyrenaicum	1	Mazandaran, SangDeh Forest	36°01ʹ11ʺ	53°13ʹ26ʺ	1720	HSBU 201,666
G. pyrenaicum	2	Mazandaran, Veresk Bridge	35°54ʹ12ʺ	52°59ʹ27ʺ	1579	HSBU 201,667
G. pyrenaicum	3	East Azerbaijan Kaleybar	38°52ʹ39ʺ	47°23ʹ92ʺ	1144	HSBU 201,668
G. pyrenaicum	4	East Azerbaijan Kaleybar cheshme Ali-Akbar	38°52ʹ39ʺ	47°25ʹ92 ʺ	1133	HSBU 201,669
G. pyrenaicum	5	East Azerbaijan Kaleybar, Shojabad	38°52ʹ39ʺ	47°25ʹ92ʺ	1137	HSBU 201,670
G. pyrenaicum	6	East Azerbaijan Kaleybar, road side	38°52ʹ39ʺ	47°25ʹ92 ʺ	1122	HSBU 201,671
G. pyrenaicum	7	East Azerbaijan, Babak fort	38°51ʹ51ʺ	47°02ʹ28ʺ	1155	HSBU 201,672
G. pusillum	1	East Azerbaijan Kaleybar	38°52ʹ39ʺ	47°23ʹ92ʺ	1144	HSBU 201,673
G. pusillum	2	Mazandaran, SangDeh Forest	36°01ʹ11ʺ	53°13ʹ26ʺ	1720	HSBU 201,674
G. pusillum	3	East Azerbaijan Kaleybar cheshme Ali-Akbar	38°52ʹ39ʺ	47°25ʹ92ʺ	1133	HSBU 201,675
G. pusillum	4	East Azerbaijan Kaleybar, Shojabad	38°52ʹ39ʺ	47°25ʹ92 ʺ	1137	HSBU 201,676
G. pusillum	5	East Azerbaijan Kaleybar, road side	38°52ʹ39ʺ	47°25ʹ92ʺ	1122	HSBU 201,677

Figure 1. Distribution map of the populations studied.

Morphological and micromorphological studies

In total 80 morphological (42 qualitative, 38 quantitative) characters were studied (Table 2). Ten plant specimens were randomly studied for morphological analyses. In each location 10 individuals were studied and examined for pollen morphology (five qualitative and eight quantitative features) (Table 4). For SEM studies, the pollen grains and seeds were directly transferred by a fine pipette to a metallic stub using double sided adhesive tape and then coated with gold in a sputtering chamber (Sputter Coater (SCDOOS, BalTec, Switzerland)). Coating with gold by the physical vapor deposition method (PVD) was restricted to 100 Å. SEM examination was carried out on a TESCAN, VEGA, SB, EasyProbe(Czech Republic). The terminology of Hesse et al. (2009) for pollen sculptures was followed. In order to study the fruit and seed morphology, 10 individuals of each location were examined for 11 qualitative and eight quantitative features (Tables 5 and 6) under a handheld digital stereomicroscope, Dino-Lite Pro (AM413T Model (www.dinolite.us)) and scanning electron microscope (SEM, JSM-6380 (JEOL, Tokyo, Japan)). The terminology used is in accordance with previously published studies of Lawrence (1970), Radford et al. (1974) and Punt et al. (1994).

Table 2. Evaluated morphological characters.

NO	Characters	Numerical code	No.	Characters	Numerical code
1	Plant height	mm	20	Mericarp length	mm
2	Length of stem leaves petiole	mm	21	Mericarp width	mm
3	Length of stem leaves	mm	22	Mericarp length/mericarp width	mm
4	Width of stem leaves	mm	23	Seed length	mm
5	Length of stem leaves/width of stem leaves	mm	24	Seed width	mm
6	Width of stem leaves/length of stem leaves	mm	25	Seed length/seed width	mm
7	Number of segment stem leaves	mm	26	Stipules length	mm
8	Length of basal leaves petiole	mm	27	Stipules width	mm
9	Length of basal leaves	mm	28	Stipules length/stipules width	mm
10	Width of basal leaves	mm	29	Bract length	mm
11	Length of basal leaves/width of basal leaves	mm	30	Bract width	mm
12	Width of basal leaves/length of basal leaves	mm	31	Bract length/bract width	mm
13	Number of segment basal leaves	-	32	Pedicel length	mm
14	Calyx length	mm	33	Peduncle length	mm
15	Calyx width	mm	34	Rostrum length	mm
16	Calyx length/calyx width	mm	35	Style length	mm
17	Petal length	mm	36	Stamen filament length	mm
18	Petal width	mm	37	Fruit length	mm
19	Petal length/petal width	mm	38	Number of flowers per inflorescence	-
39	Type root		60	Bract shape
40	Vegetation-forms		61	Stipules shape
41	State of stem strength		62	Bract and stipules hair density
42	State of stem branches		63	Bract and stipules hair
43	Leaf shape		64	Shape of segments cauline leaves
44	Phylotaxy		65	Shape of calyx
45	Leaf tips		66	Calyx apex
46	Shape of segments basal leaves		67	Petal shape
47	Stamen filament color		68	State of petal ligule
48	Stigma hair		69	Shape of petal lobes
49	Mericarp shape		70	State of petal ligule hair
50	Mericarp surface		71	Stamen filament hair
51	Mericarp hair		72	Mericarp hair density
52	Mericarp rostrum hair		73	Mericarp color
53	Sepal hair		74	Seed color
54	Sepal hair density		75	Seed shape
55	Peduncle and pedicel hair		76	Seed surface ornamentation
56	Anther color		77	Peduncle and pedicel hair density
57	Stem hair		78	Petioles hair
58	Stem hair density		79	Petioles hair density
59	Leaf hair		80	Leaf hair density

Table 3. The main differences between G. pusillum and G. pyrenaicum.

Characters	G. pusillum	G. pyrenaicum
Length and width of corolla	2–3.5 mm× 1.1–1.7 mm	5–8 mm× 2–5 mm
Length and width of leaf	1–2 × 1.5–3.5 cm	1–3.5 × 2–6 cm
Length and width of mericarp	1.7–2 mm× 0.9–1 mm	2.5–3 mm× 1–1.3 mm
Seed length	1.5–1.7 mm	2–2.2 mm
Plant height	9–30 cm	30–60 cm
Vegetation-forms	Annual or biennial	Perennial
Hair of stem and petioles	Hair of glandular and non-glandular, short fine, coarse	Hair of glandular and non-glandular, short and tall fine, Coarse
Hair of sepal, peduncle and pedicel	Long hair and medium non-glandular and glandular	Hair of glandular and non-glandular, short fine, Coarse
Shape of petal lobes	Apex emarginate with a 0.2–0.5 mm notch	Apex emarginate with a 1–2.5 mm notch

Table 4. Evaluated characters of pollen grains in studied Geranium species.

characters	G. pusillum	G. pyrenaicum
Polar axis length (p)	39.65	64
Exine thickness	3.7	4.3
Equatorial axis length (E)	38.22	59
Apocolpium length	26–27	33
Mesocolpium length	24	30
Colpus length	15.66	23
Pore diameter	7.87	4
P/E ratio	1.02	1.08
Polar shape	Circular	Circular
Equatorial shape	Circular	Circular
Homogeneity of pollen sculpture	Present	Present
Pollen shape	Spheroidal	Prolate-spheroidal
Exine ornamentation type	Reticulate-cristatum-clavate	Reticulate-cristatum-clavate

Table 5. Fruit characters of the Geranium species.

Taxon	Mericarp shape	Mericarp surface	Mericarp hair	Mericarp color	Mericarp ornamentation	Mericarp length (mm)	Mericarp width (mm)	L/W Mericarp (mm)	Rostrum length (mm)	Fruit length (mm)
G. pyrenaicum	Suborbicular	Smooth	Short hairs of non-glandular type	Brown	Reticulate	2.4–3	1–1.3	2–2.7	10–12	15
G. pusillum	Suborbicular	Smooth	Short hairs of non-glandular type	Brown	Reticulate	1.7–2	0.9–1	1.7–2.1	5–8	8–12

Table 6. Seed characters of the Geranium species.

taxon	Seed color	Seed shape	Seed surface	Seed length (mm)	Seed width (mm)	L/W seed (mm)	Hilum	Cell shape	Anticlinal wall
G. pyrenaicum	Reddish brown	Narrow elliptical	Reticulate-foveate	2–2.2	0.8–1	2.2–2.4	Basal	Polygonal	Shallow, thick, undulate
G. pusillum	Reddish brown	Narrow elliptical	Reticulate-foveate	1.5–1.7	0.8–1	1.6–2.1	Basal	Polygonal	Shallow, thick, undulate

DNA extraction and ISSR assay

Fresh leaves were used randomly from 5–10 plants in each of the studied populations. These were dried by silica gel powder. CTAB activated charcoal protocol was used to extract genomic DNA (Sheidai et al. 2013). The quality of extracted DNA was examined by running on 0.8% agarose gel.

Ten ISSR primers were used: (AGC) 5GT, (CA) 7GT, (AGC) 5GG, UBC 810, (CA) 7AT, (GA) 9C, UBC 807, UBC 811, (GA) 9T and (GT) 7CA commercialized by UBC (University of British Columbia). PCR reactions were carried in a 25 μl volume containing 10 mM Tris-HCl buffer at pH 8; 50 mM KCl; 1.5 mM MgCl₂; 0.2 mM of each dNTP (Bioron GmbH, Germany); 0.2 μM of a single primer; 20 ng genomic DNA and 3 U of Taq DNA polymerase (Bioron). The amplification reactions were performed in Techne TC-PLUS thermal cycler (GmbH, Germany) with the following program: 5 min initial denaturation step 94°C, followed by 40 cycles of 1 min at 94°C; 1 min at 52–57°C and 2 min at 72°C. The reaction was completed by final extension step of 7–10 min at 72°C. The amplification products were observed by running on 1% agarose gel, followed by ethidium bromide staining. The fragment size was estimated by using a 100 bp molecular size ladder (Fermentas Life Science, St. Leon-Rot, Germany).

Data analyses

Morphological and micromorphology studies

Morphological and micromorphological characters were first standardized (mean = 0, variance = 1) and used to establish Euclidean distance among pairs of taxa (Podani 2000). For a grouping of the plant specimens, The UPGMA (unweighted paired group using average) and Ward (minimum spherical characters) as well as ordination methods of MDS (multidimensional scaling) and PCoA (principal coordinate analysis) were used (Podani 2000). PCA (principal components analysis) biplot was used to identify the most variable morphological characters among the studied populations (Podani 2000). PAST version 2.17 (Hammer et al. 2012) was used for multivariate statistical analyses of morphological data.

Molecular analyses

ISSR bands obtained were coded as binary characters (presence = 1, absence = 0) and used for genetic diversity analysis. Genetic diversity parameter of Nei’s gene diversity (H), Shannon information index (I), number of effective loci, and percentage of polymorphism (Weising et al. 2005; Freeland et al. 2011) were determined by GenAlex 6.4 (Peakall and Smouse 2006). Nei’s genetic distance among populations was used for neighbor joining (NJ) clustering and neighbor-net networking (Huson and Bryant 2006; Freeland et al. 2011). Mantel test checked the correlation between geographical and genetic distance of the studied populations (Podani 2000). These analyses were done by PAST ver. 2.17 (Hammer et al. 2012), Darwin ver. 5 (2012) and SplitsTree4 V4.13.1 (2013) software. AMOVA (analysis of molecular variance) test (with 1000 permutations) as implemented in GenAlex 6.4 (Peakall and Smouse 2006), and Nei’s Gst analysis as implemented in GenoDive ver.2 (2013) (Meirmans and Van Tienderen 2004) were used to show genetic difference of the populations. Moreover, populations’ genetic differentiation was studied by G’ST est = standardized measure of genetic differentiation (Hedrick 2005), and D_est = Jost measure of differentiation (Jost 2008).

The genetic structure of populations was studied by 2 methods: STRUCTURE analysis (Pritchard et al. 2000)and K-means clustering (Meirmans 2012). We used two summary statistics to present K-Means clustering: 1-pseudo-F (Calinski & Harabasz 1974) and 2- Bayesian Information Criterion (BIC) (Schwarz 1978). The highest pseudo-F is used to provide the best fit clustering while in BIC, the lowest is used (Meirmans 2012). Gene flow was determined by (i) calculating Nm, an estimate of gene flow from Gst by PopGene ver. 1.32 (1997), as: Nm = 0.5(1 – Gst)/Gst. This approach considers an equal amount of gene flow among all populations. (ii) Population assignment test based on maximum likelihood as performed in Genodive ver. in GenoDive ver. 2. (2013). The presence of shared alleles was determined by drawing the reticulogram network based on the least square method by DARwin ver 5. (2012).

Results

Species delimitation in G. pusillum and G. pyrenaicum

Morphometry analysis

In the studied specimens we did not encounter intermediate forms, but the plants in each species did show some degree of morphological variability and overlap. Different clustering and ordination methods produced similar results; therefore, only MDS plots of both quantitative and qualitative morphological characters are presented here (Figure 2). Both methods separated G. pusillum and G. pyrenaicum plants in two different major groups.

Figure 2. Multidimensional scaling plots of morphological characters, separating two Geranium species groups from each other.

PCA analysis revealed that the first three components comprised about 70% of total morphological variability. In the first PCA components with about 47% of total variation, characters like length and width of mericarp, seed length, shape of petal lobes and hair of stem and petioles had the highest positive correlation (> 0.80). Length and width of corolla, plant height, length and width of stem leaves, length and width of basal leaves had the highest positive correlation (> 0.70) with the second PCA component. These characters separated G. pusillum and G. pyrenaicum from each other. Details of morphological difference in two studied species are provided in Table 3.

Pollen morphology

Details of pollen characteristics in two studied species are provided in Figure 3 and Table 4. Detailed description of pollen features in the studied species are as follows.

Figure 3. Pollen micrographs of Geranium species: G1–G3: G. pyrenaicum; H1–H3: G. pusillum. G1, H1: equatorial view; G2, H2: polar view; G3, H3: exine sculpture.

Exine ornamentations were found to be of reticulate-cristatum-clavate type, apertures were tricolporate, radially symmetric, and equatorial view was circular. Although the form of pollen grains was generally spheroidal-prolate, G. pusillum was only spheroidal (Figure 3). Although there was no significant difference between the species studied in terms of pollen grains, it was found that the grains of G. pyrenaicum were larger than G. pusillum.

These characters could separate G. pyrenaicum and G. pusillum from each other. This result is in agreement with our morphometric results.

Seed and fruit micromorphology

Details of seed and fruit micromorphology characteristics in two studied species are provided in Figure 4, and Tables 5 and 6. Detailed description of seed and fruit micromorphology features in the studied species are as follows.

Figure 4. Scanning electron micrographs (SEM) of seed and fruit coat in Geranium species. (A1–A2) (C1–C2) G. pusillum; (B1–B2), (D1–D2) G. pyrenaicum.

The mericarp surface was smooth, with a longitudinal rib, pilose, with appressed-eglandular hairs, with a few ciliae at the base in G. pusillum, while these cilia are not present at G. pyrenaicum. Mericarp and seeds are larger in G. pyrenaicum than G. pusillum.

These characters could separate G. pyrenaicum and G. pusillum from each other. This result is in agreement with our morphometric and pollen morphology results.

Intra-specific variation of G. pyrenaicum

The MDS plot of G. pyrenaicum populations based on morphological characters (Figure 5) showed morphological difference/divergence among most of the studied populations. PCA analysis revealed that the first four components comprised about 74% of total variation. Morphological characters such as petiole length in basal leaf, leaf length, and leaf width had the highest correlation (> 0.70) with the first PCA axis that comprised 33% of total variation. The length and width of sepal and corolla were the most variable characters of PCA axes 2 and 3 respectively.

Figure 5. MDS plot of morphological characters in G. pyrenaicum populations studied. Different colors indicate the plant specimens (numbers 1–8) studied from each geographical population.

Genetic diversity parameters determined in seven geographical populations of G. pyrenaicum are presented in Table 7. The highest value of polymorphism percentage (62.30%) occurred in the Mazandaran, Sang-Deh Forest population, that also had a high value for gene diversity (0.22) and Shannon information index (0.33). The East Azerbaijan, Kaleybar, Babak fort population has the lowest value for the percentage of polymorphism (3.28%) and the lowest value for Shannon information index (0.017), and gene diversity (He = 0.011).

Table 7. Genetic diversity parameters in the studied populations. (N = number of samples, Ne = number of effective alleles, I = Shannon information index, He = gene diversity, UHe = unbiased gene diversity, P% = percentage of polymorphism, populations).

Pop	N	Na	Ne	I	He	UHe	%P
pop1	7.000	1.311	1.377	0.331	0.221	0.238	62.30%
pop2	5.000	1.066	1.268	0.233	0.155	0.172	45.90%
pop3	5.000	0.443	1.059	0.057	0.037	0.041	11.48%
pop4	8.000	0.885	1.236	0.209	0.139	0.148	39.34%
pop5	8.000	0.639	1.148	0.129	0.086	0.092	24.59%
pop6	11.000	0.984	1.202	0.194	0.125	0.131	42.62%
pop7	5.000	0.311	1.020	0.017	0.011	0.012	3.28%

Populations 1–7 are according to Table 1.

Significant molecular difference (P = 0.01) was obtained among the studied population by AMOVA that was supported by G’st analysis (0.562, P = 0.001). Moreover, significant values for Hedrick standardized fixation index after 999 permutation (G’st = 0.342, P = 0.001) and Jost’ differentiation index (D_est = 0.145, P = 0.001), indicate that the geographical populations of G. pyrenaicum are genetically differentiated.

MDS plots of ISSR data (Figure 6) separated the studied populations from each other. The plot also showed the presence of a higher within-population genetic diversity in the populations of Mazandaran, Veresk Bridge, and East Azerbaijan, Kaleybar, cheshme Ali-Akbar.

Figure 6. MDS plot of ISSR data in G. pyrenaicum populations studied.

Mantel test after 5000 permutations produced significant correlation between genetic distance and geographical distance in these populations (r = 0.55, P = 0.0002). Therefore, the populations that are geographically more distant have less amount of gene flow, and we have isolation by distance (IBD) in G. pyrenaicum.

The comparison between genetic identity and genetic distance (results not shown) showed a genetic similarity (0.89) between populations of East Azerbaijan, Kaleybar, cheshme Ali-Akbar and East Azerbaijan, Kaleybar, road side, while the lowest genetic similarity value (0.76) occurs between Mazandaran, Veresk Bridge and East Azerbaijan, Babak fort populations.

The neighbor-net diagram (Figure 7) also revealed that we have almost complete separation of the studied population in the network, supporting AMOVA result. Populations 1, 2 and 5, 6 are distinct and stand separate from the other populations with great distance. Populations 3, 4 and 4, 7 show closer genetic affinity and are placed close to each other. The K-means clustering result showed that the best clustering (optimum number of genetic groups = k) according to Calinski & Harabasz’ pseudo-F was k = 6 (the highest value of pseudo-F = 8.554). The optimum number of k according to BIC was 7 (the lowest value of BIC = 867.9). Similar result was obtained by Evanno test performed on STRUCTURE analysis which produced a major peak at k = 6. STRUCTURE plot based on k = 6 (Figure 7), revealed that populations 1, 2, and 6 are distinct in their genetic content (were differently colored), while populations 3 and 4, and populations 4 and 7 had some degree of shared alleles. This genetic grouping is in agreement with the neighbor-net diagram result presented before. Both these analyses revealed that Geranium pyrenaicum populations show genetic stratification. The reticulograms obtained revealed gene flow/shared alleles among most of the studied populations (results not shown). The mean Nm = 0.39 was obtained for all ISSR loci, which indicates low amount of gene flow among the populations and supports genetic stratification as indicated by neighbor-net diagram and STRUCTURE analyses.

Figure 7. STRUCTURE plot and neighbor-net network of G. pyrenaicum populations based on k = 6 of ISSR data.

A consensus tree obtained based on ISSR and morphological trees (results not shown), revealed population 4, that is diverged from the other populations based on both morphological and molecular features. A detailed comparison of the characteristics in this population revealed that population 4 has the longest pedicle length (8–9 mm), the narrowest petal width (4–5.7 mm), the longest petal length (8.7–9 mm), the narrowest sepal width (1.7mm), and the longest sepal length (4.7 mm).

Intra-specific variation of G. pusillum

ANOVA test revealed a significant difference in quantitative morphological characters among the studied populations (P < 0.05). Different cluster analysis and ordination methods performed based on morphological characters produced similar results; therefore, only the MDS plot is presented and discussed (Figure 8). The results showed morphological difference/divergence among most of the studied populations (Figure 8). This morphological difference was due to quantitative characters only. PCA analysis of morphological characters revealed that the first three PCA components comprised about 71.5% of total variation. Morphological characters, e.g. the length and width of bract, length and width of sepal, length of peduncle, length of pedicle, the width of seed and mericarp, had the highest value of correlation with these components and are the most variable morphological characters among the studied populations.

Figure 8. MDS plot of morphological characters in G. pusillum populations studied. Different colors indicate the plant specimens (numbers 1–8) studied from each geographical population.

Genetic diversity parameters determined in five geographical populations of G. pusillum are presented in Table 8. The highest value of percentage polymorphism (56.86%) was observed in East Azerbaijan, Kaleybar, cheshme Ali-Akbar which shows high value for gene diversity (0.164) and Shannon information index (0.25). The East Azerbaijan, Kaleybar, road side population has the lowest value for the percentage of polymorphism (9.80%) and the lowest value for Shannon information index (0.051), and He (0.033).

Table 8. Genetic diversity parameters in the studied populations. (N = number of samples, Ne = number of effective alleles, I = Shannon information index, He = gene diversity, UHe = unbiased gene diversity, P% = percentage of polymorphism, populations).

Pop	N	Na	Ne	I	He	UHe	%P
pop1	5.000	0.922	1.190	0.173	0.114	0.127	33.33%
pop2	6.000	0.843	1.168	0.156	0.102	0.112	31.37%
pop3	9.000	1.275	1.261	0.256	0.164	0.173	56.86%
pop4	9.000	1.137	1.292	0.267	0.176	0.186	52.94%
pop5	6.000	0.471	1.054	0.051	0.033	0.036	9.80%

Significant molecular differences (P = 0.01) are highlighted by the AMOVA test, with 53% and 47%, respectively, as inter- and intra-population genetic variability. These results indicate that the geographical populations of G. pusillum are genetically differentiated from each other.

Non-metric MDS plots of ISSR data (Figure 9) showed both intra- and inter-population genetic diversity and revealed close genetic affinity between populations 3 and 4. The plots were higher within population genetic diversity in the populations East Azerbaijan, Kaleybar, cheshme Ali-Akbar and East Azerbaijan, Kaleybar, Shojabad. The studied populations were placed in a separate group which was in agreement with the AMOVA result. Mantel test after 5000 permutations produced significant correlation between genetic distance and geographical distance in these populations (r = 0.55, P = 0.0002). Therefore, the populations that are geographically more distant have less gene flow, and we have isolation by distance (IBD) in G. pusillum.

Figure 9. MDS plot of ISSR in G. pusillum populations studied. Different colors indicate the plant specimens (numbers 1–8) studied from each geographical population.

The comparison between genetic identity and genetic distance (results not shown) showed a genetic similarity (0.89) between populations East Azerbaijan, Kaleybar, cheshme Ali-Akbar and East Azerbaijan, Kaleybar, Shojabad, while the lowest genetic similarity value (0.63) occurs between East Azerbaijan, Kaleybar and Mazandaran, SangDeh Forest.

The NJ tree (results not shown) produced two major clusters. East Azerbaijan, Kaleybar, cheshme Ali-Akbar and East Azerbaijan, Kaleybar, Shojabad showed close genetic affinity and formed the first cluster. Mazandaran, Sang-Deh Forest was placed in a single cluster far from the other studied populations. The NJ tree supported the grouping made by the MDS plot (Figures 8 and 9).

The STRUCTURE plot based on k = 3 was performed with admixture model ancestry and revealed a genetic affinity between East Azerbaijan, Kaleybar, cheshme Ali-Akbar and East Azerbaijan, Kaleybar, Shojabad (Figure 10(b)). This is in agreement with NJ tree and MDS plot presented before. Both these analyses revealed that G. pusillum populations show genetic stratification.

Figure 10. (a) Reticulogram, (b) STRUCTURE plot, (c) neighbor-net network of G. pyrenaicum populations based on k = 3 of ISSR data.

The neighbor-net diagram (Figure 10(c)) also revealed genetic similarity between populations 3, 4 and 1, 5. supporting the STRUCTURE plot. The reticulogram obtained (Nm = 0.32; Figure 10(a)) supported a low degree of gene flow/ancestral shared sequences between 4 and 5, as well as between 4 and 1, 2. This result is in agreement with the STRUCTURE plot of the studied species based on ISSR nuclear data and all these results are in agreement in showing high degree of genetic stratification within G. pusillum populations.

A consensus tree was obtained for both ISSR and morphological trees (Figure 11), to reveal the populations 1, 2 and 4 that are diverged based on both morphological and molecular features. Detailed comparison of the characteristics in these populations revealed that, for example, population 4 has the longest sepal length (3.5 mm), the longest sepal width (1.5 mm), the narrowest seed width (0.7 mm) and the narrowest mericarp width (0.9 mm), populations 1 has the narrowest peduncle length (7.5 mm), the narrowest bract length (1.4 mm), and the narrowest bract width (0.2 mm), population 2 has the longest seed width (1 mm), and the narrowest fruit length (8 mm).

Figure 11. Consensus tree of morphological and molecular data in G. pusillum populations.

Discussion

Species delimitation and taxonomic consideration

The species delimitation in complex groups and in those in which the species have different degree of morphological overlap is a tedious and difficult task. In these situations, it is suggested to use different and combined approaches, including morphological, molecular, and cytological methods, to determine the species boundaries (Carstens et al. 2013). In the last few decades the use of molecular markers as tools for species and subspecies delimitation has drastically increased (Sheidai et al. 2013). The basic premise for the use of molecular markers for species delimitation is that the “species tree” should be inferred from a “gene tree”. A previous study on taxonomic revision of Geranium sections Batrachioidea and Divaricata Rouy. (Geraniaceae) was performed (Aedo et al. 1998b). According to Van Loon (l984c), intraspecific crosses were often highly successful in this section, but the only successful interspecific cross was that involving G. molle and G. brutium. According to (Aedo et al. 1998b) G. luganense, G. oenense, and G. hybridum are probably not hybrids but synonyms of G. molle (the first two) or G. pusillum. Therefore, we performed morphological and molecular study of G. pusillum and G. pyrenaicum of the section Batrachioidea to study the species inter-relationships. These species often occur together. In the studied specimens we did not encounter intermediate forms.

In the specimens studied by Esfandani Bozchaloyi et al. (2017b), an MDS plot based on both quantitative and qualitative morphological characters delimiting the species studied in sect. Batrachioide and did not encountered with intermediate forms, which is in agreement with the present results.

The previous studies showed the importance of fruit micromorphology in taxonomy of the genus. Fruit characters were found to be important for separating taxa at infra-generic level but were not efficient to differentiate the sections (Keshavarzi 2015; Salimi Moghadam et al. 2015). However, seed morphology was found to be useful in the taxonomic delimitation at the generic, specific as well as at the infra-specific levels (Ather et al. 2012). Similarly, the pollen morphology is significantly useful at the generic level (Perveen and Gaiser 1999).

UPGMA clustering and TCS networking of the studied populations based on ISSR markers did not entirely delimit the studied species and revealed that plants in these species are inter-mixed. In the UPGMA circular dendrogram, a higher degree of intermixture occurred between G. pusillum and G. pyrenaicum (Esfandani Bozchaloyi et al. 2017b).

UPGMA trees of the combined dataset of ITS and rbcL supported separation of the three species as their accessions formed separate clusters with high bootstrap value (> 0.98). In general UPGMA trees of cp-DNA and ITS trees are in agreement with morphology data. They showed affinity between G. pusillum and G. pyrenaicum. They also separated G. molle from the other species (Esfandani Bozchaloyi et al. 2017b).

The present study revealed that in micro-morphological study only quantitative features are of diagnostic importance in Geranium species studied. The species relationship obtained is also in agreement with morphological analysis and supports taxonomic treatment of Flora Iranica (Schönbeck-Temesy 1970).

Intra-specific morphological and genetic variability

Population genetic study provides valuable information about the genetic structure of plants, the stratification versus gene flow among the species populations, genetic divergence of the populations, etc. (Esfandani Bozchaloyi et al. 2017b). This information has different applications, and from pure understanding of the biology of the species to the conservation of endangered species, choosing of proper parents for hybridization and breeding and phylogeography and mechanism of invasion (Freeland et al. 2011). Geranium pyrenaicum is widespread in Iran and it has several medicinal applications (Baytop 1999); however, we had no information on its genetic structure and detailed taxonomic information. The present study revealed interesting data about its genetic variability, genetic stratification and morphological divergence throughout the country. The degree of genetic variability within a species is highly correlated with its reproductive mode; the higher degree of open pollination/cross breeding brings about higher level of genetic variability in the studied taxon (Freeland et al. 2011). G. pyrenaicum and G. pusillum are mainly self-pollinating species; therefore, a low level of genetic variability within populations in these species might be related to the closer nature of breeding in this taxon.

Another well-known feature of self-pollinating species is high among-population genetic and morphological divergence. This happens due to a limited amount of gene flow or its complete absence among geographical population in a single species (Freeland et al. 2011). The present study also revealed significant morphological and genetic difference among G. pyrenaicum and G. pusillum populations, quite in agreement with the mentioned assumption. This is particularly supported by the STRUCTURE plot that identified separate genetic groups within this population and by consensus tree of both morphological and genetic data. Different mechanisms like isolation, drift, founder effects and local selection may act to bring about among population differentiation, and therefore, populations differ in phenotypic traits and allelic composition (Jolivet and Bernasconi 2007). We should state that the studied populations differed in quantitative morphological characters and we do not know how much of the morphological difference among the studied populations is genetically controlled; they may be under influence of environmental conditions. Therefore we do not attempt to suggest new taxonomic forms below the species level for these taxon and consider them as different ecotypes only.

The present population divergence may be under influence of isolation by distance across the distribution range of the studied populations. The dispersal of these populations might be constrained by distance and gene flow is most likely to occur between neighboring populations. As a result, more closely situated populations tend to be more genetically similar to one another (Slatkin 1993; Hutchison and Templeton 1999; Medrano and Herrera 2008). The populations’ divergence may be accompanied by local adaptation. When we use multilocus molecular markers (such as SSR, AFLP, RAPD, and ISSR) for population genetic studies we understand that these are neutral molecular markers (they are not directly acting as adaptive genes), but they may be linked to a gene or a genetic region with adaptive value (Freeland et al. 2011).

Acknowledgments

The authors wish to thank Saeed Javadi Anaghizi in Central Laboratory of the Shahid Beheshti University for providing SEM pictures. We are indebted Prof. Carlos Aedo, Madrid, Spain, for his generous help in commenting on the manuscript. The authors thank anonymous reviewers for valuable comments on an earlier draft.

Disclosure statement

No potential conflict of interest was reported by the authors.