Characterization of Valuable Indigenous Barberry (Berberis sp.) Germplasm by Using Multivariate Analysis

ABSTRACT

Seedless barberry (Berberis vulgaris L. var. asperma) is cultivated in arid and semi-arid areas of Iran. It is widely used as a food additive. Thirty-nine morphological characteristics were evaluated to estimate the genetic variability among 42 barberry accessions during 2010–2012. Analysis of variance showed that all of the variables tested in examined accessions were significant (P ≤ 0.05), with a high level of variability. There were significant negative and positive correlations among some of the characteristics of the accessions. Ten major factors that accounted for 79% of variation were identified. Cluster analysis divided the accessions into five major groups based on all of variables analyzed. The three main groups were berry size, seedlessness, and vitamin C level. Using three main factors, the accessions were plotted in three dimensions; ‘R2N1’ had larger berries with a higher dry weight, while ‘R12N2’ had smaller leaves, a less dry weight, and more thorns. Quantitative and qualitative characteristics of the berries such as color, size, and shape, the number of seeds per 100 berries, and the number of berries per bunch, taste, and acidity showed a high level of variation among the studied accessions. The high genetic diversity among barberry accessions in our collection indicates that it is a valuable resource that can be used in barberry breeding programs in the future.

KEYWORDS:

• Barberry
• factor analysis
• genetic diversity
• cluster analysis
• morphological characteristic

Introduction

Barberry, a member of Berberidaceae, has a long history of cultivation and consumption in Iran, while in the rest of the world it is cultivated as ornamental or medicinal plant. Seedless barberry is cultivated mainly in the Southern Khorasan region (East of Iran) to produce dried berries and is a major source of income for farmers in the region (Balandari, 2002; Saeed et al., 2007). The berries are also used for the production of juices, beverages, powder, jam, and marmalade, and the bark and root are used in the pharmaceutical industries. In other parts of the world, barberry is consumed more as processed fruit than as fresh fruit (Balandari, 2002; Saeed et al., 2007). However, in Iran, only during harvest season are fresh berries supplied to the local markets in Iran (Balandari, 2002).

The Berberidaceae contains 15 genera and 650 species, mostly found in the northern hemisphere. In the southern hemisphere, the genus Berberis is found in the temperate zone of South America. Berberidaceae includes evergreen and deciduous shrubs, which grow in a wide range of ecological conditions (Ahrendt, 1961; Bottini et al., 2002). Berberis are propagated by seed or vegetative parts (rhizomes, suckers, and cuttings), the rhizomes can grow to a length 10 m far from the crown, so populations with a low number of plants per hectare always have a lot of ground coverage (Ahrendt, 1961). There is considerable uncertainty regarding systematics among the genus. Some scholars divided Berberis into 500 species with simple leaves and 200 species with compound leaves. However, other researchers believe that the actual number may be less than this. Ahrendt (1961) identified 60 species of Berberis in Chile and southern Argentina, while Landrum (1999) only accepted 20 species in that collection as Berberis. Landrum (1999) divided barberries with simple leaves into 29 subgroups and those with compound leaves into 4 subgroups. However, Landrum (1999) later changed this grouping. In a study using taxonomic and phytogeography implications from internal transcribed spacer phylogeny in Berberis, Kim et al. (2004) showed that species of barberries that were previously divided into subgroups by Schneider (1905) and Ahrendt (1961) were very similar. However, there were some questions about these subgroups, especially for Berberis microphylla, Berberis buxifolia, and Berberis hetrophylla. According to these researchers, there are five species of wild barberries in Iran including Berberis vulgaris, Berberis orthobotrys, Berberis crataegina, Berberis integerrima, and Berberis khorasanica, with the last species native to Iran (Balandari, 2002; Ghahreman, 1996; Sabeti, 1997; Termeh and Matin, 1982). The genus that was cultivated in Iran as commercial barberry was given a different name. Ghahreman (1996) named this genus as B. khorasanica Browiz & Zielinski, although Sabeti (1997) introduced this species as unknown or B. sp. It also was named B. orientalis Ck Schn var. asperma and B. vulgaris L. var. asperma, but the origin of plants receiving those scientific names is uncertain (Ghahreman 1996). Although there are botanical keys and descriptors for the identification of barberry species in Iran, the primary identity of true species in most cases has been lost due to inter- or intra-cross-pollination among genera and species. Thus, their profiles are not a perfect match with available descriptors (Balandari, 2002).

According to the cultivation of this unique plant in Iran and the possibility of adaptation and cultivation of this shrub in a desert climate with cool nights and hot days, and saline water and soil, obtaining accurate information about morphological and genetic characteristics of Iranian barberries is critical. Such information can be used to establish breeding programs for developing new varieties for the fresh market, as well as medicinal and ornamental purposes. Research programs focusing on developing varieties with high quantity and quality of fruit and resistance to adverse environmental conditions will ultimately create appropriate cultivars to introduce globally. Description of morphological variability is useful for determining phenotypic characteristics that contribute to the total diversity in a germplasm collection. Multivariate data analysis is a powerful statistical technique for analyzing genetic variation among plant populations. Qualitative and quantitative morphological characteristics can provide a useful data set for phenotypic variation among plant accessions by multivariate analysis (De Oliveira et al., 2012; Furones-Pérez and Fernández-López, 2009; Mehmood et al., 2014). Factor analysis, cluster analysis, and tri-plot analysis are the most popular multivariate techniques for characterization of accessions using morphological characteristics. Application of these techniques can provide useful information regarding morphological traits that contribute to genetic variation within germplasm collections (Khodadadi et al., 2011; Mohammadi and Prasanna, 2003). Multivariate analysis methods have been applied by many researchers to evaluate genetic variation in some fruit tree germplasm collections (Ganopoulosa et al., 2016; Khadivi-Khub, 2014; Sarkhosh et al., 2009). However, we have found no studies whereby multivariate analyses have been used to differentiate accessions in an indigenous barberry germplasm collection in Iran.

With respect to developing commercial barberry varieties for different markets globally, this study was conducted in order to gain more knowledge about relationships among characteristics of Iranian barberry accessions. To the best of our knowledge, this is the first report on the evaluation of quantitative and qualitative morphological traits of Iranian Berberis accessions.

Materials and methods

Plant materials

The 42 barberry (Berberis spp.) accessions studied originated from the germplasm collection at the Technology and Science Park, Mashad, Iran (Table 1). Each accession in the collection is represented by three 6-year-old (mature) trees; all three trees were used to collect the data during 2010–2012 (2 years). The germplasm collection is located 36°26^׳05^״ N latitude and 985 m above the sea level with an annual average temperature of 14.3°C and an annual average precipitation of 251 mm. Horticultural management practices including irrigation, fertilizer, and spraying for pests and diseases were done at regular intervals each year.

Table 1. List of barberry accessions used in this study with their some specific characters.

No	Genotype cod	Stigma remanding on fruit	Fruit color	Fruit shape
1	R1N1	Medium	Red	Obovate
2	R1N2	Short	Light red	Ovule
3	R1N3	Short	Brick red	Obovate
4	R2N1	Sticked	Red	Ovate with angle
5	R2N2	Short	Scarlet	Obovate
6	R2N3	Medium	Brick red	Ovule
7	R3N1	Medium	Light red	Ovule
8	R3N2	Short	Crimson	Obovate
9	R3N3	Short	Light red	Ovate with angle
10	R4N1	Medium	Red	Obovate
11	R4N2	Medium	Crimson	Obovate
12	R4N3	Medium	Red	Obovate
13	R5N1	Short	Blackish red	Globular
14	R5N2	Medium	Brick red	Ovule
15	R5N3	Medium	Red	Obovate
16	R6N2	Tall	Light red	Small long
17	R6N3	Tall	Brick red	Ovule
18	R7N1	Medium	Red	Obovate
19	R7N2	Short	Light red	Obovate
20	R8N1	Medium	Red	Obovate
21	R8N2	Medium	Red	Obovate
22	R8N3	Short	Light red	Ovate with angle
23	R9N1	Short	Blackish red	Obovate
24	R9N3	Sticked	Bluish black	Obovate
25	R10N1	Sticked	Blackish red	Ovule
26	R10N2-1	Short	Light red	Ovule
27	R10N2-2	Tall	Light red	Small long
28	R10N3	Medium	Red	Obovate
29	R11N1	Medium	Blackish red	Globular
30	R11N2	Medium	Bluish black	Oblong
31	R11N3	Medium	Blackish red	Globular
32	R12N1	Sticked	Bluish black	Obovate
33	R12N2	Medium	Bluish black	Small long
34	R12N3	Short	Red	Obovate
35	R13N1	Medium	Red	Obovate
36	R13N2	Short	Red	Obovate
37	R14N1	Medium	Blackish red	Globular
38	R14N2	Medium	Bluish black	Globular
39	R16N1	Medium	Brick red	Globular
40	R16N2	Medium	Light red	Obovate
41	R16N3	Medium	Red	Obovate
42	Bd	Medium	Red	Ovule

Analysis of morphophysiological characteristics

Thirty-nine characteristics in barberry accessions were measured based on morphometric data and chemical analysis (Table 2). A digital caliper (Anyi Instrument, Guangxi, China) was used to measure dimensions (millimeter) such as length and diameter for leaves, berries, and seeds. Weight (g) of different tissues was measured with an electronic balance (Ohaus SP602 AM, Fotronic Corporation, MA, USA). Berry juice was used to analyze total soluble solids (TSS), titratable acidity (TA), and pH (HANNA; Woonsocket, RI, USA). TSS (Brix°) was measured with a refractometer (Pocket PAL-1 ATAGO Corporation, Tokyo, Japan). TA was measured by neutralization to pH 8.10 with 0.10 N NaOH. Data are reported as mg/l of citric acid. Also, some characteristics such as shrub shape, branching habit, thorn intensity, branch color at the end of summer, the number of suckers per shrub, and berry shape, leaf shape, seed shape, and berry and seed color were measured based on ordinal and nominal data. For measuring anthocyanins, 1 ml of berry juice was diluted in 3 ml deionized water and the absorbance of the diluted solution was measured at 510 nm with a UV-Visible spectrophotometer (Perkin Elmer, Lambda EZ201, CA, USA). Deionized water was used as a blank correction. A WinDIAS leaf area meter (Delta-T Devices Ltd, Cambridge, UK) was used to measure the leaf area index. Ascorbic acid content was measured by redox titration using an iodine solution (Verdini and Lagier, 2000). Shrub habit, shooting habit, sucker rate, shoot color, number of thorn, thorn density, stigma remanding on berry, berry color berry shape, seed shape, seed color, leaf shape, leaf color, and leaf blade density were measured, coded, or rated based on botanical descriptors available for Berberis (Balandari, 2002; Ghahreman, 1996; Sabeti, 1997; Saeed et al., 2007; Schneider, 1905; Termeh and Matin, 1982).

Table 2. Measured berry characteristics and descriptive statistics among the studied barberry accessions.

No.	Character	Abbreviation	Unit	Min	Max	Mean	CV%^z
1	Shrub habit	ShrHa	Ordinal (1–5)	1.00	5.00	3.05	23.14
2	Shooting habit	ShoHa	Ordinal (1–5)	1.00	5.00	2.76	43.53
3	Sucker rate	SuR	Ordinal (1–5)	1.00	5.00	3.05	31.07
4	Shoot color (last summer)	ShoC	Nominal (1–5)	1.00	5.00	3.27	29.84
5	Number of berry per bunch	BeNu	Number	4.67	32.33	16.41	41.96
6	Number of bunch in 1/3 of end shoots	BuNu	Number	0.33	19.67	6.05	58.10
7	Thorn number	ThoNu	Number	1.00	3.00	1.34	56.80
8	Thorn density	ThoDe	Ordinal (1–3)	3.00	5.00	3.24	20.43
9	Thorn length	ThoL	mm	5.00	30.07	10.30	34.47
10	Berry waxy	BeWa	Ordinal (0–3)	0.00	3.00	1.29	81.58
11	Stigma remanding on berry	BeSt	Ordinal (1–7)	1.00	7.00	3.83	42.07
12	Berry color	BeC	Nominal (1–13)	1.00	13.00	7.34	44.72
13	Berry shape	BeSh	Nominal (1–11)	1.00	13.00	9.00	34.78
14	Berry length	BeL	mm	6.20	14.50	10.90	15.26
15	Berry diameter	BeDi	mm	4.20	8.70	6.46	15.37
16	Berry length/diameter ratio	BeLDi	Ratio	0.90	2.26	1.71	15.67
17	Beesh berry weight	FBeW	g	8.47	55.60	21.46	40.73
18	Dry berry weight	DBeW	g	3.00	12.60	6.56	39.42
19	Berry dry mass percentage	BeDrP	%	20.00	48.28	31.00	23.67
20	Berry juice percent	BeJ	cc	13.33	53.33	31.04	30.32
21	Total soluble solution	TSS	%	9.50	26.00	19.91	19.51
22	Total titration acidity	TA	%	2.68	8.44	5.19	40.02
23	Berry flavor index	BeFI	Ratio	2.16	8.61	4.28	37.90
24	Anthocyanin	Anto	OD510nm	0.001	3.33	0.89	97.16
25	Acidity	pH	pH	2.16	3.80	2.81	10.70
26	Vitamin C	VitC	mg/100g	29.33	76.27	45.76	25.60
27	The seed number in berry	SNuBe	Number	0.01	183.33	53.37	70.70
28	Seed shape	SSh	Nominal (1–11)	3.00	13.00	9.05	34.41
29	Seed color	SCo	Nominal (1–7)	1.00	9.00	4.71	40.93
30	Seed length	SL	mm	40.60	70.80	60.33	12.09
31	Seed diameter	SDi	mm	10.33	40.00	20.22	18.88
32	Seed length/diameter ratio	SLDi	Ratio	1.95	4.75	2.93	20.81
33	Seed weight	SW	g	0.55	1.40	0.83	22.16
34	Leaf shape	LeSh	Nominal (1–9)	1.00	9.00	5.34	38.28
35	Leaf blade shape	LeBSh	Nominal (1–9)	1.00	9.00	4.22	49.57
36	Leaf blade density	LeBDe	Ordinal (1–5)	1.00	7.00	4.12	40.67
37	Leaf length	LeL	mm	20.29	40.86	30.79	15.41
38	Leaf width	LeWi	mm	0.99	2.50	1.86	15.27
39	Leaf area index	LeAI	mm^2	24.66	84.17	55.96	23.48

^zCoefficient of variation = (standard deviation/mean) * 100.

Statistical analyses

The experiment was arranged as a randomized complete block design and data were analyzed by analysis of variance (ANOVA) and mean treatment differences were tested at P < 0.05 by an least significant difference (LSD) test using SAS version 9.3 statistical software (SAS Institute, 2000). The following values were evaluated for all variables: mean, minimum, maximum, and coefficient of variation (CV%). A correlation coefficient for parametric and non-parametric characteristics was obtained by Pearson and Spearman methods, respectively (Mohammadi and Prasanna, 2003). Multivariate ANOVA, factor analysis (Varimax rotation), and clustering of genotypes (Ward’s method) using mean values of each parameter was used to estimate relationships among accessions. SPSS statistical software ver. 10 (SPSS Inc., Chicago, USA, Norusis, 1998) was used for multivariate and non-parametric statistical tests. Also, a tri-plot was formed according to the Factor 1, 2, and 3 (F₁, F₂, and F₃).

Results

Analysis of variance

The results of ANOVA showed significant differences among genotypes for all traits (P ≤ 0.05). Mean values and ranges of variability for the different characteristics among accessions are presented in Table 2. The coefficient of variation (CV) ranged from 10.70% to 97.16%. Characteristics with a high CV indicate a large variety that can be used for selection (Mohammadi and Prasanna, 2003). Characteristics such as the number of berries per bunch, number of bunches in one-third of the end shoots, berry anthocyanin levels, and berry fresh weight had considerably high CVs (Figure 1). The lowest CV (10.70%) was for juice acidity (pH), whereas anthocyanin (97.16%) and berry waxy (81.58%) had the highest CVs. Twenty-three characteristics exhibited a CV greater than 30%, indicating a high level of variation in those morphophysiological traits among the studied accessions (Table 2 and Figure 1). Berry characteristics are important traits that influence appeal to consumers for the fresh market. The accessions ‘R7N1’ (14.5 mm) and ‘R1N3’ (6.2 mm) had the greatest and least berry length, respectively. Berry diameter was largest in ‘R2N1’ (8.7 mm) and smallest in ‘R10N2-2’ (4.7 mm). The greatest average weight of 100 berries was 55.6 g for the ‘R2N1’ accession, although significant variation was observed among accessions for this characteristic CV = 40.37% (Table 2). The number of berries per bunch varied remarkably among accessions, from a minimum of 4.67 (‘R11N3’) to a maximum of 32.33 (‘Bd’). There were significant differences in berry dry weights % among accessions with a range of 20% (‘R10N3’) to 48.28% (‘R11N2’). The CV value for berry dry mass percentage was 23.67% (Table 2).

Figure 1. Diversity among barberries accessions: (a) commercial orchard of seedless barberry in Iran, (b) seedless barberry fruit at time of harvesting, (c) fruit variation among the 42 accessions, (d) seedless barberry at the full bloom stage, (e) berry variation among some accession, and (f) variation among leaves for some studied accessions.

In berries of barberry the juice, pulp, seed, and other compounds were evaluated. TSS (Brix°) is one of the most important characteristics. The highest TSS value was 26% (‘R11N1’) and the lowest was 9.5% (‘R10N1’) with a CV of 19.51%. Because both sour and sweet accessions were considered in this study, a high variability was detected for TA among the accessions, which ranged from 2.68 (‘R2N3’) to 8.44 (‘R13N1’). There was a wide range of berry juice pH values from 2.16 (‘R10N2-1’) to 3.8 (‘R9N3’). The berry flavor index (TA/TSS) varied from 2.16 (‘R3N3’) to 8.61 (‘R6N3’). The highest value for berry juice was 53.33% (‘Bd’), and the lowest value was 13.33% (‘R1N1’). The highest (3.33) and lowest (0.001_560nm) anthocyanin levels were for ‘R12N2’ and ‘R3N3’, respectively. There was a significant difference among accessions for the vitamin C (mg 100 g⁻¹) with a range of 29.33 mg 100 g⁻¹ (‘R1N1’) to 76.27 mg 100 g⁻¹ (‘R5N1’).

Correlations among characters

Results of simple correlation analysis indicated the existence of significant positive and negative correlations among measured parametric (Table 3) and non-parametric characteristics (Table 4). Relationships among fruit size, peel color, and fruit chemical traits have been reported for some other fruit crops (Khadivi-Khub, 2014, Sarkhosh et al., 2009). A high positive correlation was observed among Shrub habit and sucker rate and shoot color (r = 0.45 and 40, P ≤ 0.01, Table 3), lying shrub accessions produced more suckers and had a darker shoot color. A significant negative correlation (r = −0.45) was found between shooting habit and berry shape, indicating that shrubs with bent branches had more elongated berries than those with straighter branches. Waxy berry was negatively correlated with stigma remanding on the berry and positively correlated with berry color and leaf blade shape (Table 3). A significant negative correlation was observed between berry characteristics (length, diameter, and fresh berry weight) and anthocyanin concentration (Table 4). However, a significant positive correlation was detected between berry characteristics and leaf characteristics (leaf length, width, and leaf area index). The relationship between leaf size and berry quality traits (vitamin C content and Brix°) was very high, indicating that accessions with larger leaves had larger berries (Tables 3 and 4). Production characteristics such as the number of berries per bunch, the number of bunches per shoot, and berry juice percentage were positively correlated with each other. However, there was a negative correlation with the number of seeds per berry and juice pH. Vitamin C content and Brix were positively correlated. A similar correlation was reported by Sarkhosh et al. (2009) for pomegranate fruit.

Table 3. Bivariate correlations among the non-parametric measured characters in the studied barberry accessions.

Character◘	ShrHa	ShoHa	SuR	ShCo	ThoNu	ThoDe	BeWa	BeSt	BeC	BeSh	SSh	SCo	LeSh	LeBsh	LeBDe
ShrHa	1
ShoHa	0.14	1
SuR	0.45**	0.016	1
ShC	0.40**	0.031	0.2	1
ThoNu	−0.03	0.19	0.11	−0.11	1
ThoDe	−0.24	0.05	−0.18	−0.24	0.43**	1
BeWa	0.1	−0.31*	0	0.15	0.16	0.09	1
BeSt	−0.22	0.2	0	−0.16	−0.26	−0.31*	−0.52**	1
BeCo	0.01	−0.37*	−0.04	0.1	0.13	0.1	0.48**	−0.49**	1
BeSh	−0.39**	−0.45**	−0.16	−0.05	0.3	0.11	0.08	−0.14	0.03	1
SSh	0.2	0.13	0.12	0.07	0.02	−0.15	0.02	0.07	0.16	−0.41**	1
SC	0.02	−0.2	−0.17	0.3	−0.08	−0.37*	−0.19	0.36*	−0.24	0.33*	−0.26	1
LeSh	0.13	−0.02	0.1	0.33*	−0.06	−0.26	0.18	−0.33*	−0.08	0.23	−0.15	0.12	1
LeBsh	0.18	−0.04	0.03	−0.04	0.1	−0.16	−0.04	0.02	0.41**	−0.18	0.26*	0.14	−0.08	1
LeBDe	−0.16	−0.15	−0.32*	−0.06	0.06	0.2	0.23	−0.21	0.56**	0.03	−0.03	−0.22	−0.21	0.3	1

* Significant at level of 5%.

** Significant at level of 1%.

Table 4. Bivariate correlations among the parametric measured characters for the studied barberry accessions.

◘	BeNu	BuNu	ThoL	BeL	BeDi	BeLDi	FBeW	DBeW	FDryP	BeJ	TSS	Anto	pH	TA	VitC	BeFI	SNuBe	SL	SDi	SLDi	SW	LeL	LefW	LwAI
BeNu	1
BuNu	0.53**	1
ThoL	−0.16	−0.29	1
BeL	0.074	−0.07	0.22	1
BeDi	0.052	−0.22	0.07	0.52**	1
BeLDi	−0.01	0.17	0.14	0.44**	−0.5**	1
FBeW	0.012	−0.16	0.1	0.67**	0.82**	−0.21	1
DBeW	−0.17	−0.39*	0.12	0.65**	0.73**	−0.24	0.83**	1
BeDP	−0.26	−0.37*	−0.07	−0.06	−0.24	0.24	−0.24	0.28	1
BeJ	0.24	0.46**	−0.03	−0.08	0.18	−0.34*	0.16	−0.15	−0.67**	1
TSS	0.19	−0.21	0.09	0.32*	0.31*	−0.03	0.28	0.49**	0.38*	−0.12	1
Anto	−0.14	−0.052	−0.28	−0.5**	−.41*	0.02	−0.33*	−0.09	0.57**	−.36*	0.11	1
pH	−0.49**	−0.29	−0.1	−0.15	−0.03	−0.11	−0.07	0.12	0.28	−0.2	−0.17	0.33*	1
TA	0.24	−0.15	0.25	0.37*	0.2	0.1	0.33*	0.39*	0.08	−0.001	0.34*	−0.22	0.09	1
VitC	−0.15	−0.36*	0.05	0.13	0.03	0.05	0.15	0.39*	0.45**	−0.31*	0.54**	0.27	0.2	0.257	1
BeFI	−0.07	−0.05	−0.15	−0.21	−0.031	−0.18	−0.18	−0.1	0.15	−0.07	0.34*	0.31*	−0.21	−0.7**	0.05	1
SNuBe	−0.40**	−0.44**	0.01	0.05	0.11	−0.07	0.09	0.52**	0.66**	−0.7**	0.04	0.37*	0.45**	0.18	0.41**	−0.2	1
SL	−0.04	0	0.32*	0.50**	−0.14	0.62**	0.13	−0.01	−0.05	−0.02	−0.117	−0.28	−0.06	0.32*	0.03	−0.52**	−0.02	1
SDi	−0.05	−0.08	0.21	0.05	0.11	−0.09	0.05	0.04	0	0.01	−0.26	−0.03	0.17	−0.07	0.1	−0.12	0.22	0.17	1
SLDi	−0.03	0.04	0.11	0.29	−0.23	0.46**	0.022	−0.08	−0.04	0.02	0.13	−0.18	−0.17	0.29	−0.11	−0.26	−0.24	0.49**	−.73**	1
SW	−0.19	−0.12	−0.05	0.37*	0.19	0.146	0.27	0.17	0.05	−0.01	−0.21	−0.23	0.29	0.09	0.02	−0.43**	0.3	0.54**	0.41**	0.05	1
LeL	0.17	−0.14	0.29	0.46**	0.296	0.09	0.47**	0.44**	−0.12	0.11	0.52**	−0.3	−0.13	0.28	0.27	0.07	−0.19	0.26	−0.07	0.24	−0.11	1
LeWi	0.19	−0.18	0.16	0.33*	0.49**	−0.22	0.53**	0.47**	−0.11	0.07	0.57**	−0.16	−0.15	0.08	0.24	0.27	−0.09	−0.1	−0.25	0.12	−0.22	0.74**	1
LwAI	0.23	−0.12	0.14	0.27	0.33*	−0.11	0.31*	0.34*	−0.05	0.02	0.60**	−0.21	−0.1	0.16	0.18	0.27	−0.1	−0.11	−0.12	0.05	−0.28	0.78**	0.81**	1

* Significant at level of 5%.

** Significant at level of 1%.

Factor analysis

Table 5 shows the results of the factor analysis. The variance for each factor indicates the contribution, as a percentage, of that factor to the total variance of due to all measured variables. The first tens factors with eigenvalue ≥1 accounted for 79% of the total variance. The characters with threshold levels greater than 0.5 were considered significant for each factor (Table 5). The first factor accounted for 11.36% of the overall variance. Variables such as thorn density, total soluble solution, leaf length, leaf width, and leaf area index had the highest factor loading. Factor 1 included leaf size, TSS, and thorn density. In the second factor (F₂), the largest scores were due to traits associated with the berry such as berry color, berry dry weight percent, berry juice percentage, total titration acidity, pH, and the number of seeds per berry. High loading characters in F₃ included berry length, berry length:diameter ratio, seed shape, seed length, seed weight, and leaf blade shape. According to Table 4, other factors were grouped as berry factor (F₄), berry, leaf, seed, shoot, thorn factors (F₅, F₆, F₇, and F₉), and shrub shape and seed (F₈ and F₁₀).

Table 5. Eigenvalue accepted (≥1), variance and cumulative variance for 10 factors results from factor analysis.

Factors		1	2	3	4	5	6	7	8	9	10
Eigenvalue		4.43	3.99	3.97	3.78	3.39	2.65	2.6	2.36	1.91	1.72
% of variance		11.36	10.23	10.17	9.69	8.7	6.79	6.68	6.06	4.9	4.42
Cumulative variance		11.36	21.6	31.77	41.46	50.16	56.95	63.63	69.69	74.59	79
Characteristics	Unit	Factor loading*
Shrub habit	Ordinal (1–5)	−0.06	2.02	0.19	0.17	−0.09	0.12	0.08	0.77*	0.15	−0.14
Shooting habit	Ordinal (1–5)	−0.19	−0.16	0.15	0.02	−0.84*	0.02	0.06	0.06	−0.05	−0.15
Sucker rate	Ordinal (1–5)	0.26	−0.06	0.1	−0.3	0	−0.01	0.17	0.5	0.27	−0.51*
Shoot color (last summer)	Nominal (1–5)	0.01	0.08	−0.03	0.12	0.06	−0.27	0.04	0.76*	−0.01	0.18
Number of berry per bunch	Number	0.31	−0.29	0.08	−0.13	−0.4	0.26	0.02	0.19	−0.57*	−0.01
Number of bunch in 1/3 of end shoots	Number	−0.07	−0.44	0	−0.28	0.35	0.01	0.05	0.03	−0.53*	−0.19
Thorn number	Number	−0.19	0.05	0.01	−0.13	−0.21	−0.23	0.64*	0.23	0.26	−0.01
Thorn density	Ordinal (1–3)	−0.61*	0.13	−0.21	0.08	−0.12	−0.17	0.22	−0.44	0.15	−0.15
Thorn length	mm	0.17	−0.07	0.24	−0.01	−0.1	0.16	0.11	0.2	0.70*	0.11
Berry waxy	Ordinal (0–3)	−0.37	0.22	0.17	−0.16	0.53*	−0.44	0.1	0.16	0.02	−0.04
Stigma remanding on Berry	Ordinal (1–7)	0.37	−0.03	0.06	−0.38	−0.37	0.51*	−0.35	−0.14	0.04	0.16
Berry color	Nominal (1–13)	−0.2	0.58*	0.27	−0.02	0.51*	−0.22	0.31	0.02	−0.11	−0.06
Berry shape	Nominal (1–11)	0.24	−0.23	−0.41	0.06	0.41	−0.18	0.27	−0.19	−0.03	0.50*
Berry length	mm	0.28	−0.01	0.58*	0.56*	−0.17	0.03	−0.1	0.22	0.04	−0.16
Berry diameter	mm	0.22	−0.21	−0.16	0.87*	0.05	0.02	0.11	0.11	0.05	−0.1
Berry length/diameter ratio	ratio	−0.02	0.23	0.72*	−0.35	−0.26	0	−0.23	0.1	−0.01	−0.07
Fresh berry weight	g	0.24	−0.13	0.15	0.89*	−0.05	0.08	0	0.08	−0.02	0.08
Dry berry weight	g	0.27	0.31	−0.01	0.81*	0.05	0.16	0.01	0.06	0.09	0.05
Berry dry mass percent	%	0.02	0.86*	0.02	0.08	0.19	0.01	−0.07	0.1	−0.01	−0.09
Berry juice percent	CC	0.05	−0.72*	−0.09	0.03	−0.02	0.04	0.08	0.08	−0.14	−0.06
Total soluble solution	%	0.70*	0.4	−0.13	0.22	−0.18	0.08	−0.15	0.16	0.01	−0.13
Total titration acidity	%	−0.1	0.66*	0.22	−0.29	0.12	−0.16	0.03	0.02	−0.27	−0.05
Berry flavor index	Ratio	−0.11	0.23	−0.07	−0.03	0.75*	0.19	0	0.03	0.01	−0.16
Anthocyanin	OD510nm	0.16	0.07	0.18	0.23	0.02	0.86*	−0.09	0.02	0.07	−0.02
Acidity	pH	0.32	0.61*	−0.05	0.09	0.05	0.28	0.11	0.2	0.03	0.11
Vitamin C	mg/100g	0.31	0.19	−0.37	−0.13	−0.18	−0.76*	0	0.03	0	−0.06
The seed number in berry	Number	−0.15	0.58*	0.08	0.3	0.41	0	0.27	−0.16	0.07	0.05
Seed shape	Nominal (1–11)	0.07	−0.06	0.77*	0.11	−0.02	0.187	0.06	−0.12	0.1	−0.24
Seed color	Nominal (1–7)	0.33	0.04	0.18	−0.13	−0.1	0.06	−0.18	0.17	0.27	0.78*
Seed length	mm	0.03	−0.07	0.90*	0.03	0	0.16	0.06	0.11	0.11	0.02
Seed diameter	mm	−0.12	−0.03	−0.16	0.06	0.14	0.04	0.87*	0.23	0.02	−0.02
Seed length/diameter ratio	Ratio	0.09	−0.08	0.42	−0.12	−0.05	0.04	−0.79*	−0.05	0.15	0.09
Seed weight	g	−0.25	−0.08	0.56*	0.28	0.46	0	0.11	0.26	−0.15	0.02
Leaf shape	Nominal (1–9)	0.18	−0.42	−0.07	0.29	0.11	0.51*	−0.03	0.34	−0.09	0.18
Leaf blade shape	Nominal (1–9)	0.08	0.32	0.53*	0.05	0.08	0.06	0.28	0.05	−0.01	0.39
Leaf blade density	Ordinal (1–5)	−0.23	0.46	0.21	−0.01	0.08	−0.1	0.23	−0.16	−0.53*	0.2
Leaf length	mm	0.84*	0.09	0.25	0.2	−0.07	0.1	−0.06	−0.09	0.12	0.08
Leaf width	cm	0.80*	0.05	−0.07	0.37	0	−0.14	−0.17	0.01	0.02	0.12
Leaf area index	mm2	0.91*	−0.07	−0.08	0.13	0.02	−0.08	0	0.05	0.1	0.06

*Significant factor loading (value above 0.50).

Cluster analysis

Cluster analysis based on all morphological characters, grouped the accessions into five subgroups with a respective distance of 15 out of 25 (Figure 2). Group I included 17 accessions with similar characteristics such as large, heavier berries, high vitamin C content, high juice acidity (TA), large leaves, red berries, waxy berries, low anthocyanin content, and poor berry flavor index (TSS/TA) that are reported in Tables 1 and 2. Also, all accession except ‘R3N3’ had light green seed color and elongated seeds. In general, based on botanical characteristics of wild barberries in Iran (Balandari, 2002; Ghahreman, 1996; Sabeti, 1997; Termeh and Matin, 1982), accessions located in this group are similar to B integerrima, indicating that they probably belong to this species or its hybrids. Group II consisted of nine accessions with relatively large leaves, lower TA, high TSS, low anthocyanin content, free of a waxy layer on the berry, light seed color, few seeds per berry, a large number of berries per bunch, and large number of bunches per shoot. The accessions (‘R3N1’, ‘R5N2’, ‘R6N3’, and ‘R2N3’) that are cultivated on a commercial scale in the Khorasan Province of Iran (Balandari, 2002) were located in this group. Group III included seven accessions. Berries in this group have high TSS, high vitamin C and anthocyanin contents, and round seeds and berries. With the exception of ‘R14N1’, ‘R1N2’, and ‘R1N3’, all accessions in this group had dark red berries with a large number of seeds. Accession characteristics in this group were similar to those of B. khorasanica (Balandari, 2002; Ghahreman, 1996; Sabeti, 1997, Termeh and Matin, 1982). Therefore it seems this group belongs to B. khorasanica or are hybrids derived from it. Group IV included four accessions characterized by black berries, round leaves with a low leaf area index, black seeds, high vitamin C content, high anthocyanin content, low fresh and dry berry weight, elongated berries, and overhanging branches. Characteristics of this group are fairly similar to those of B. crataegina. Group V included five accessions characterized by a low leaf area index, low TSS, high anthocyanin content, heavy seeds, and black color berries. Also, berries in this group had a round shape with a thick layer of wax and 1-year-old shoot color was brown. According to botanical report, this group appears to be derived from B. orthobotrys (Balandari, 2002; Ghahreman, 1996; Sabeti, 1997; Termeh and Matin, 1982). In general, some differences in characteristics such as berry taste, and seed and berry shape and size had the greatest effect on separating accessions into different clusters (Figure 2).

Figure 2. Dendrogram of grouping 42 barberry accessions based using Ward’s method.

Tri-plot analysis

Tri-plot analysis characterized the relationship among accessions regrading phenotypic similarity and berry characteristics and supported the results of cluster analysis (Figure 3).

Figure 3. Tri-plot analysis of 42 barberry accessions based on three main factors (F₁, F₂, and F₃, Table 5).

Discussion

Genetic studies and variety characterization of barberry are valuable for use in breeding programs with the goal of introducing new promising commercial cultivars. Different methods based on morphological and molecular markers have been used to differentiate varieties. Evaluation of barberry accessions is important not only for genotyping, but also to record and quantify the characteristics that might be useful in improving barberry according to breeding objectives. Variability among fruit characteristics is mainly influenced by the accession. However, climatic conditions and agronomic practices may play an essential role (White et al., 2012). ANOVA revealed that there is significant variation among the accessions (P ≤ 0.5). This study recognized that some accessions have individual characteristics that are suitable to use for cultivation or in breeding programs for developing promising cultivars for different proposes such as fresh market, and the pharmaceutical and beverage industries. Understanding the relationships among different traits would help plant breeders quantify complex characteristics, thus saving time and expense. The correlation coefficients for particular traits can provide information that is valuable in assessing accessions in germplasm collections. Instead of directly measuring traits that are costly or difficult to measure, the variability of such a trait can be measured indirectly by measuring another trait that is highly correlated with it (Norman et al., 2011).

Factor analysis revealed that berry and seed characteristics were dominant in the first five factors and contained most of the total variance (50.16%). These characteristics varied the most among the studied accessions. Attending to the extensive amount of data obtained from assessment of different morphological characters in a germplasm collection can be unmanageable. Therefore, by applying factor analysis, different characteristic can be grouped from with other highly correlated characteristic. With this analytical method, it is much easier for researchers to select a few of the many available characteristics (Dennis and Adams 1997; Mohammadi and Prasanna, 2003). In this study, berry characteristics exhibited the highest variation and can be used to evaluate of barberry accessions. Relationships between characteristics exposed by factor analysis may correspond to genetic linkages among loci of controlling characteristics or pleiotropic effects (Iezzoni and Pritts, 1991). In the present study, factor analysis grouped the variables into 10 main groups of which the first and second accounted for a high proportion of the overall variance (Table 5). Cluster analysis provided a useful and comprehensive tool to establish a first order of genotypic classification. Grouping based on several characteristics can be a reliable method to determine the similarities and differences among accessions (White et al., 2012). Discriminant analysis showed a significant difference among the groups, indicating that these groups were created relatively far from each other based on morphophysiological characteristics, where berry shape, berry color, berry dry weight, berry length, and stigma remanding on berry were the most important characteristics for differentiating groups. Also, the grouping of barberry accessions in this study was supported by researchers, who have discriminated barberry species by morphological characteristics in previous studies (Balandari, 2002; Ghahreman, 1996; Sabeti, 1997; Termeh and Matin, 1982). Tri-plot analysis generated a two or three consequent descriptions that either of consequent can be a main discrepancy factor. Scattering of accessions at the area of this main factor discriminates variances among accessions (Mohammadi and Prasanna, 2003). In this study, the percentage of relative variance between the first four factors was approximately 10%. Thus, the cumulative variance for first three factors total was 31.77% (Table 5). Because of no significant difference in the relative variance between first four factors used to assess the studied accessions, factors 1, 2, and 4, which seem are important in pomology and breeding programs, were used to perform tri-plot analysis. Factors F1 (leaf characteristics), F2 (qualitative berry characteristics), and F4 (quantitative berry characteristics) described 31.28 of the total variance.

Conclusion

Multivariate analysis revealed genetic variability in morphological characteristics among the barberry accessions evaluated. The present study provided the most inclusive morphological analysis to date among barberry accessions. For instance, 100 fresh berries weigh from 8.47 to 55.6 g, berry length is from 6.20 to 14.5 mm, juice percentage ranges from 13.33% to 53.33% of the berry, berry shape is from round to elongated, and seed number per berry varies from 0 to 3.66 (Table 2). TSS ranged from 9.5% to 26%; anthocyanins from 0.1 to 3.33; and vitamin C levels from 29.33 to 76.27 mg 100 ml⁻¹. A comprehensive morphological data set including berry quality, quantity, and vegetative characteristics is a compulsory information for a germplasm collecting to allow breeders to hand pick appropriate parents for future breeding programs for improving berry quality and developing DNA markers for marker-assisted selection (DiStefano et al., 2013; Gmitter et al., 2012). This information will be accessible to the breeders worldwide to select appropriate accessions as parents in future breeding programs.

Acknowledgments

We thank the Research Institute for Food Science and Technology (RIFST), Mashhad, Iran, for providing access to the barberry germplasm collection.