Patterns of phenotypic variation in a germplasm collection of pepper ( Capsicum annuum L . ) from Turkey

Phenotypic diversity was assessed for quantitative and qualitative traits in a collection of pepper (Capsicum annuum L.) germplasm from different areas of Turkey. A total of 48 genotypes, including germplasm lines and commercial cultivars, were studied under field conditions at Izmir over two consecutive summers (2004 and 2005). All accessions were characterized for 67 agro-morphological traits from seedling emergence to crop maturity. Morphological data were subjected to principal components analysis followed by hierarchical agglomerative clustering. This provided seven groups based on morphological and agronomic properties. The first six principal components axes accounted for 54.29% of the variance among the 48 accessions and their lines. The greater part of variance was accounted for by traits such as fruit diameter, fruit weight, volume, fruit wall thickness, fruit productivity, and fruit soluble solid and dry matter content. The high diversity found in the collection showed its great potential for improving agronomic traits in pepper. Additional key words: cluster analysis, genetic diversity, morphological characterisation, plant genetic resources.


Introduction
Pepper (Capsicum spp.) is one of the world's major vegetable and spice crops (Zewdie et al., 2004).Csillery (2006) indicates that the first competent description of Capsicum was given in Hungarian Herbal by Dioszegi and Fazekas (1807), who used the nomenclature of Linnaeus to describe C. annuum (paprika, the Turkish pepper) and C. sinense (later C. chinense; the Chinese pepper).Pepper belongs to the family Solanaceae and includes 30 known species (Bosland and Votava, 2000).It is thought to originate from South America and has spread throughout the world, including the tropics, subtropics and temperate regions (Pickersgill, 1997).Peppers are tender perennials when grown in their native habitats.Of the 30 species included in the Capsicum genus, only five -C.annuum, C. baccatum C. chinense, C. frutescens, and C. pubescens -have been domesticated and cultivated.
Pepper is a very important crop in Turkey, where it has been cultivated for centuries; both hot and sweet varieties are grown.The country produces 410,000 tonnes of bell and 1,340,000 tonnes of long green peppers with an annual increase in production of about 4-10%.Ninety percent of the peppers produced are marketed fresh; the remaining 10% are processed (Abak, 1994).The most important processed forms of pepper are pepper paste, paprika and spice.Red sweet pepper is mainly used in the paste industry, while very hot and sweet peppers are used in the spice industry.The southern and southeastern cities of Turkey are the most important sources of red pepper for spices (Abak, 1994).However, it is not clear how peppers found their way into the country.Andrews (1999) suggests several possibilities, although the most accepted hypothesis is that in the 15 th and 16 th centuries they reached Istanbul from Portuguese East Africa and India, arriving via Arab Middle Eastern trade routes.
Turkey is one of the most important nations in the world for pepper genetic resources, and the large numbers of cultivars grown around the country provide an important source of variation for plant breeding.A number of accessions have been collected from different regions of Turkey (Anonymous, 2007) for this very reason, but no systematic study has ever been undertaken to investigate the extent of genetic variation nor the relationships between different Turkish pepper genotypes.
Most pepper cultivars currently grown in Turkey are open pollinated.Some local pepper landraces are still grown on many small farms due to consumer demand.
For decades these have been cultivated in different environments and using different growing techniques.In general, they are genetically diverse and well adapted to the locations where they have been developed (Votava et al., 2005).Although pepper plants are considered autogamous (Allard, 1960), high rates of cross pollination (7% to 90%) have been recorded in several studies, and cross pollination events could change the genetic identity of these landraces.Further, given the importance of commercial pepper production in Turkey, many growers have turned away from the traditional cultivars towards new, higher yielding cultivars that produce fruit suitable for processing.Certainly, hybrid varieties are used for greenhouse production.
Estimating genetic diversity and determining the relationships between germplasm collections helps ensure germplasm is efficiently collected and managed.Data on the level of genetic diversity of a germplasm collection may also increase the efficiency of efforts to improve a species (Geleta et al., 2005).Plant breeders can use genetic similarity information to complement phenotypic information in the development of breeding populations (Nienhuis et al., 1993;Greene et al., 2004;Yüzbaşıoğlu et al., 2006).The total seed protein content, isoenzyme profiles, and several types of molecular marker can be used to determine the variability and relationships among accessions (Rabbani et al., 1998), and, indeed, Capsicum species has been analysed using morphological, cytogenetic and molecular markers (employing restriction fragment length polymorphism [RFLP], random amplified polymorphic DNA [RAPD], and amplified fragment length polymorphism [AFLP] techniques) (Conicella et al., 1990;Lefebvre et al., 1993Lefebvre et al., , 2001;;Zewdie and Zeven, 1997;Geleta et al., 2004).
However, morphological characterisation is the first step in the description and classification of germplasm (Smith and Smith, 1989).The main aim of this study was to analyse the morphological and agronomic traits of Turkish pepper accessions collected from different parts of the country in order to assess their genetic diversity.

Material and methods
The experimental material included samples of 48 landraces and cultivars of pepper grown in Turkey (Table 1): 30 accessions from the Aegean Agricultural Research Institute (AARI) (collected from different regions), 14 local cultivars (both hot and sweet types widely used for fresh consumption and processing), and four cultivars from the US Chile Pepper Institute (sweet and hot peppers) (Table 1 The experimental design was a randomised complete block with three replicates; each plot consisted of 20 plants.Data were collected from 10 plants.Five unharvested plants were left to collect data regarding full fruit maturity, and five were left to reduce side effects.Each accession was evaluated.Some accessions showed very strong variation among genotypes, thus, in the first year 48 genotypes were examined but in the second 94 genotypes and germplasm lines were recognised and examined.Soil preparation, fertilization and plant protection were undertaken following the usual practices for pepper in Turkey (Vural et al., 2000).
Data were collected on 67 morphological and physiological traits defined by the International Board for Plant Genetic Resources Descriptors for Capsicum (IPGRI, AVRDC and CATE, 1995) and more recent investigations (AOAC, 1995;Ngouajio et al., 2003;Gibbs and O'Garo, 2004) (Table 2).
Genotype characteristics were recorded as quantitative or qualitative values as required.The methodology used to record qualitative values from seedling to harvest was obtained from the descriptor for Capsicum (IPGRI, AVRDC and CATE, 1995) (Table 2) (Zewdie and Zeven, 1997).Principal component analysis (PCA) was performed on all variables.Hierarchical agglomerative clustering was then performed on the principle component axes obtained, using the Ward criterion (Sneath and Sokal, 1973).This was preferred because it tends to produce compact clusters (Zewdie and Zeven, 1997).Within-cluster means and standard deviations of quantitative variables were calculated for ease of interpretation.All calculations were performed using STA-TISTICA software (Statsoft Inc., 2004).

Results
The quantitative and agronomic traits assessed showed wide variation.Among the agronomic traits, all the pepper genotypes examined had white hypocotyls with no pubescence.The cotyledon colour ranged from light to dark green.All had a yellow corolla except for TR 62374.No spots or stripes were seen on the corolla, the shape of which was rotate and campanulate.Stem colour was green for all genotypes; no anthocyanin was visible in the internodes or on the anther and stigma, except in TR 69723.No male sterility was seen nor were anthocyanin spots or stripes observed on fruit.The range of variation for most morphological traits was very large.For example, fruit shape ranged from pointed to sunken, and pointed fruit colour varied from lemon-yellow to red.The number of flowers per axil was usually one, only TR 62374 and TR 69068 had either one or two flowers per axil.These accessions are used for pickling.
PCA was used to examine the variation of the pepper genotypes.The first six axes accounted for 54.29% of the variability among the 48 accessions and their lines.Figure 1 provides a dendrogram for the studied accessions.The first axis was mainly related to variation in fruit diameter, fruit weight, fruit volume, edible fruit rate, wall thickness, and fruit soluble solid and dry matter contents (Tables 2, 3).The second axis was mainly concerned with pedicel length, fruit length and pH.The remaining eight axes were related to other fruit and plant traits (Table 3).The high total variance explained by the first three axes was shown in a 2D and 3D screen plot; each cultivar is plotted based on its principal components score (the cumulative proportion of variance) for each of the first three axes (Figs. 2 and 3).
To determine the hierarchical similarity among genotypes, a dendogram of genetic distance was made using the PCA data employing the Ward criterion (Fig. 1).Seven groups were obtained, mainly based on fruit shape and fruit agronomic traits.

Group A
This group contains 22 genotypes clustered into two subgroups.All these genotypes were obtained from the AARI, except for 'Acı süs' and 'Acı şahnalı', which are local cultivars.Group A fruits are used for either fresh consumption or processing, such as pickling or making hot sauce.The average fruit diameter of this group is 4.57 cm; fruits are small and narrow (average 1.66 cm) and the mean fruit weight is 6.28 g (Table 4).Fruit volume varies from 1.8 to 25.4 cm 3 .Fruit colour is mainly dark green.The earliest flowering genotypes (56.8 days) belong to group A. The plants of least height (55.52 cm) but highest capsaicin content (62.67 mg 100g -1 ) also fall into group A. Compared to other groups, yields are moderate.

Group B
Group B contains 15 genotypes clustered into two subgroups.All four foreign genotypes fell into this group, as did the local cultivar 'Yunan biberi', which in Turkish means Greek pepper.This seed is imported by international seed companies but it is strange that this genotype should fall in with foreign cultivars in the cluster analysis.The peppers of this group are mainly produced for the fresh market and are used in salads.The group B genotypes mainly produce either large or small block type fruit and have low capsaicin content.Although the dry matter content is low (9.56%), the yield is high (18160.3kg ha -1 ).The genotypes that belong to this group have the largest fruit volume (70.41 cm 3 ), the highest edible fruit rate (74.15%), and have a very dark colour.The genotypes with the lowest soluble solid content (6.10%) also belong to group B. Days to maturity is longer than in the other groups.The plants in this group have fruits with a thin wall (1.89 cm) and have a long fruit bearing period (99.36 days).The total soluble solids content in the juice was measured using a Atago refractometer (Tokyo, Japan) 60 pH Fruit juice acidity was measured using a Mettler Toledo MP220 pH meter (Giessen, Germany) 61 FER Edible fruit rate (%): the whole fruit was weighed, then the removed seeds and peduncle and the edible portions weighed separately.The percentage difference between whole fruit weight and that of the edible portion was then calculated.62 DMC Fruit dry matter content (%): the fruit pedicel was removed and dried in an oven at 65ºC until weight loss between measurements was <0.05 g.The percentage difference between the fresh and dry weights was used to calculate the dry matter content of the fruit.

Group C
This contains four genotypes from the AARI clustered into two subgroups, one with only one genotype.The fruits of group C are used for fresh consumption and pickling.On average these genotypes have fruits of diameter 2.07 cm, a weight of 14.07 g, a volume of 30.42 cm 3 , a wall thickness of 1.89 cm, and a soluble solid content of 6.62%.The mean plant stem length of group C genotypes is 8.51 cm, and the leaves are large (width 4.38 cm, length 8.03 cm).In addition, the fruits are long (9.30cm), dark green, the pedicel quite long (3.88 cm), and the leaves dark green in colour.The capsaicin content is 47.45 mg/100 g -Group C is the second placed group for capsaicin content.

Group D
Group D contains eight genotypes, all from the AARI, clustered into two subgroups.These genotypes are used for dry pepper production (the fruits are mainly ground).Turkey's climate is very suitable for sun-drying in the summer period, especially in southwest Anatolia.Group D genotypes are grown over large areas in this region.The fruit wall is thin (1.74 cm) and the mean fruit dry matter content is 14.20%.The highest capsaicin content is seen in this group (117.83mg/100 g); capsaicin content is affected by genotype, environment (high temperatures and water deficit increase fruit capsaicin content), growing season, cultivation practices and fruit maturity stage.Thick-walled peppers take longer to sun-dry, and some of the outer skin can peel off.The members of group D also showed a high edible fruit rate (61.68%).The fruits are dark-light green colour.

Group E
Group E consists of 10 genotypes clustered into two subgroups.They are used at the green stage for salads or "dolma" (a traditional Turkish food prepared by stuffing peppers with rice, onion and other ingredients).Ripe, red ripe fruits are suitable for paste production (these peppers develop a good red colour).Fruit yield is high.The fruits are narrow, have a low fruit volume, and have a high soluble solid and dry matter content.The plants of this group have a characteristic compact growth, a long stem with a large diameter, and a long fruit bearing period.The highest vitamin C content (131.33 mg/100 g) is seen in this group.

Group F
Group F includes six genotypes clustered into two subgroups.The fruit wall is 3.61 cm thick.These fruits have a large diameter and the highest fruit volume.These genotypes also have the highest edible fruit rate (74.34%), a soluble solid content of 6.55%, a very short 'days to flowering' value (56.97 days), but a longer 'days to maturity' value (84.04 days).They produce compact plants with long, wide leaves.The yield is some 16,850 kg ha -1 .The pungency of the fruit is medium.The titratable (0.17) acidity of these green-red fruits is low.

Group G
This group included 29 genotypes clustered into three subgroups.The second and third subgroup includes genotypes from the AARI collection.The genotypes in the first subgroup are all local cultivars except TR 69724-5 and TR 69128.The fruit of these genotypes are moderately pungent.Plants are quite tall (74.52 cm) but the fruit volume low (21.43 cm 3 ).The average soluble solid content is 6.73%.This mean dry matter content was the highest recorded (14.3%).Mean yield (17,340 kg ha -1 ) and edible fruit rate (67.63%) are also high.

Discussion
The assessment and description of trait variation is important in the initiation of programs aimed at the selection of genotypes providing high yields and which have qualitative traits acceptable to consumers.The assessment of genetic variability among genotypes is useful for the conservation of genetic resources, for broadening the genetic basis of cultivars, and for cultivar protection (Yüzbaşıoğlu et al., 2006).The degree of genetic diversity among cultivated crops depends on their reproductive behaviour (Geleta et al., 2005).Lefebvre et al. (1993)  also suggest this may be related to its reproductive behaviour.The association between genetic similarity and geographic distance among landraces is not always clear (Sonnante and Pignone, 2007).Geographical origin probably contributes to the genetic variability among the genotypes studied (Geleta et al., 2005).
In general, qualitative traits may be expected to contribute to the clustering of genotypes, but the quantitative traits are of most interest here given their importan-ce in improvement programs (Lotti et al., 2007).The variation found in quantitative traits is useful for developing varietal descriptors and in variety identification.Since quantitative traits are of agronomic interest they are of much interest in improvement programs (Panthee et al., 2004).
In the present study, the most representative variables for describing the phenotypic diversity of the genotypes were defined by PCA (Sneath and Sokal, 1973).Cluster analysis was then performed to establish groups; due to low Eigen vector values it would have been difficult to group genotypes based solely on the PC axes obtained.Lotti et al. (2007) reported no evident or significant groups among melon genotypes and variables transformed into new co-ordinates in a multi-dimensional space represented by six principal component axes.Peppers are classified into different commercial varieties based on fruit traits (Greenleaf, 1986;Geleta et al., 2005).In this study, pepper genotypes with similar fruit characteristics clustered together.This study also investigated the genetic variability and relationships among the clusters of these pepper resources.The dendogram obtained consisted of seven groups and a number of subgroups resulting from different morphological and agronomic traits.The range of variation for most morphological traits was large, inclu-ding fruit wall thickness, fruit capsaicin content, and vitamin C content; these are affected by genotype, environment, growing season, growing practice and fruit maturing stage (Lindsay and Bosland, 1996;Martinez et al., 2005).The level of variation found in the present collection shows there to be very high potential for developing pepper varieties for different processing purposes such as for drying, making pepper paste and hot sauce, capsaicin extraction, and pickling.Zewdie and Zeven (1997) report very large variation among Yugoslavian hot pepper accessions, and indicate their fruit size to range from small and circular to large and bell shaped.Fruit colour also ranged from red to yellow, growth habit ranged from prostrate to erect, and plant height from short to tall.Similar observations were made in the present work.The present study shows that the peppers distributed over the wide range of geogra-  62.67 23.58 42.01 24.77 47.45 22.50 117.83 33.49 33.42 18.60 42.41 42.73 42.53 21.95 Table 4. Means and standard deviations of traits used in the classification of each of the seven pepper clusters phic conditions of Turkey show significant variation in terms of most of their morphological traits.Indeed, many of the lines observed showed properties different to those of their mother plants.The greater part of the variation was accounted for by the fruit diameter, fruit weight, fruit volume, fruit wall thickness, edible fruit rate, and the soluble solid and dry matter contents.Cluster analysis grouped together accessions with greater morphological similarity, as reported by Zewdie and Zeven (1997) who examined variation among hot pepper accessions.These authors indicate the distribution produced by cluster analysis in their work to be mainly a consequence of the number of fruits per plant, fruit weight, and 1000 seed weight.
Estimating genetic diversity and determining the relationships between collections are very useful strategies for ensuring efficient germplasm collection and management.Different markers, including the total seed protein content, isozyme profiles and several types of molecular markers, are available for studying variability among accessions (Rabbani et al., 1998).Several techniques have been used to classify and measure the patterns of phenotypic diversity in the relationships of species and germplasm collections for a variety of crops.However, morphological characterisation is the first step in the description and classification of germplasm.Further information can then be obtained using DNA markers and molecular techniques.Geleta et al. (2005) described that both morphological traits and AFLP markers generally separate pepper genotypes according to fruit traits, and a significant positive correlation between the morphological data and AFLP marker-based matrices indicates that AFLP distances tend to reflect morphological distances.Lefebvre et al. (2001) indicated that relationships between molecular distances and phenotypic distances show that inbred lines with different phenotypes also differ in terms of their AFLP markers.Thus, a genotype can be easily discriminated with the use of phenotypic distances only (Geleta et al., 2005).
In conclusion, this work shows that Turkish pepper genotypes can be divided into seven groups based on their morphological and agronomic traits.The analysis of variance carried out on these agronomic and morphological properties showed considerable morphological variation among pepper genotypes, a consequence of the introduction of different pepper genetic material to Turkey since the 16th century (Andrews, 1999).Some genotypes that are interesting in terms of their capsaicin, dry matter and ascorbic acid contents, as well as for their fruit morphology, shape, and suitability for processing, are highlighted.The material investigated in this study indicates Turkey be very rich in pepper germplasm.Advantage could be taken of this diversity in breeding programs.

Figure 1 .
Figure 1.Dendogram for the 48 pepper genotypes/lines obtained from different regions of Turkey, produced by Ward's clusters analysis; clusters are based on morphological traits (scale: Euclidean distance).

Figure 2 .
Figure 2. Relationships among Turkish pepper genotypes shown by a 2D scatter for first two principal components (based on morphological traits).

Figure 3 .F
Figure 3. Relationships among Turkish pepper genotypes shown by a 3D scatter diagram of first three principal components (based on morphological traits).
).All experiments were conducted on a sandy-loam soil at Ege University, Faculty of Agriculture, Department of Horticulture in Bornova, Izmir Province Turkey

Table 2 .
Morphological and agronomic traits recorded in the Capsicum annuum accessions/cultivars

Table 3 .
Eigen values and proportion of variance explained by the 10 principal components with respect to 48 pepper germplasm traits