Introduction
Faba beans are one of the most grown and consumed grain food legumes in Tunisia as well as other countries in North Africa and Middle East. The field bean with small seed type (Vicia faba L. var minor), principally used for animal feeding, is considered as an alternative to soybean in the diet of dairy cows, especially in countries where soya beans have high costs of supply (Gous, 2011; Tufarelli et al., 2012). Grain legume production is declining while costs are increasing. There is also a high GMO contamination risk for soya bean (Martini et al., 2008). Field beans require the development of large collections of high quality varieties in order to offer greater choices to the market. However, research in faba beans is still facing some difficulties, especially related to genetic isolation within the genus Vicia, tolerating no exchange genes with other species of the same genus, and previous attempts at interspecific hybridization have failed (Hebblethwaite, 1983; Ramsey and Pickersgill, 1986; Le Guen and Duc, 1992; Duc, 1997; Zeid et al., 2003). Consequently, faba bean breeding is based only on natural diversity within the species or on artificial mutagenesis (Hebblethwaite, 1983; Link et al., 1995; Duc, 1997). Partial allogamy of this specie plays a role in maintaining genetic diversity, with natural outcrossing ranging from 2% to 84% and an average of 32% (Bond and Poulsen, 1983), yet it can play a negative role in maintaining pure lines and in variety conservation. For this reason, it is necessary to collect, conserve, characterize and efficiently maintain local populations (Burton, 1978; Terzopoulos et al., 2003). Terzoupolous et al. (2008) argued that it is important to study the diversity of the rich Mediterranean gene pool to identify the germplasm groups from which inbred lines can be derived. Numerous investigations have been conducted using morphological traits, allowing Polignano and Spagnoletti Zeuli (1985) and Suso et al. (1993) to classify the studied collections. Terzopouloset et al. (2003) evaluated a collection of 55 Greece populations, divided them into 5 groups and recommended the number of pods per plant, the number of ovules and seeds per pod, and the number of branches per plant as the most relevant traits to be used for population classification under dry and low fertility soils (Terzopouloset et al., 2003). Naguib (2000) evaluated morphological characters for identification of faba bean varieties by means of qualitative characters, e.g., pod color, seed coat color and hilum color along with quantitative traits including leaflet characters, number of flowers, plant height and pod characters. He concluded that these descriptors could be used for discrimination among the studied faba bean genotypes. In Tunisia, some efforts have been initiated to investigate variability in local faba beans populations (Ouji et al., 2012) and landraces (Yahia et al., 2012) based on morphology. Only, Three commercial field bean varieties (Bacchar, Badii and Najah), locally selected by the National Institute of Agricultural Research of Tunisia (INRAT), in addition to a new variety (Saber02) selected by the National Agronomic Institute of Tunis (INAT), are commercialized. However, Tunisia is one of the centers of diversity of Vicia faba L. and the available local diversity has not been adequately sampled. The aim of this study is to screen agro-morphologically a set of lines created by mass selection from an original population of field bean (Vicia faba L. Var minor). This is therefore one of the pioneering studies to characterize the INAT collection and identify lines with interesting traits. Duc et al. (2004) found that the vc-gene, which controls levels of the anti-nutritional vicine and convicine compounds, was closely linked (10 cM) to the hilum color gene. Hence; a further objective of this study is to establish whether white hilum in any of the studied lines is associated with low vicine and convicine levels. Whereas the white flower is indicative of the absence of tannins in the seed coat by a mutation of one of the two complementary gene controlling absence of tannin and white flower (Bond, 1976; Cabrera, 1988; Crofton and Bond, 1998; Gutierrez et al., 2008) were focused within this collection. This initial screening is the first step prior to carry out costly genotyping to supply INAT and International faba beans breeding programs with new sources of diversity.

1 Materials and Methods
1.1 Plant materials history
The research material involves ninety seven F4 lines of field beans (Vicia faba var. minor L.) and three registered local cultivars used as controls (Table 1). The lines originated from a local population ‘F75’ collected during the seventies which are conserved in the INAT Gene Bank. Initially, this population was sown in a large field which allowed the selection of some interesting and distinguishable individuals based on agronomical and morphological traits, complimented by protein content analysis and seed descriptors. This process led to the creation of ten sub-populations. Seeds from each sub-population were sown in separate rows in three separated randomized blocks and allowed to open pollinate. Some crossing between rows within and between blocks was inevitable. Seeds were harvested from each plant separately and given a unique identifier related to its original row and block. After this subdivision, each single plant within each line was sown in a single row in an insect-proof covered tunnel and each row was considered as a line. The current experiment entails a first screening after one cycle of purification to characterize the lines in comparison with three local cultivars.

Table 1 Codification du germoplasme Vicia faba L. var minor

 
1.2 Field experimental conditions
The field experiment was carried out at the National Agronomic Institute of Tunis (INAT), Tunis, (36°49’50’’N and 10°10’58’’W and altitude of 10 m) during the 2009/2010 growing season in a heavy clay soil. All lines and controls were sown on November 25th. Each line is represented by one row 2 m long with 1m between rows and a 35 cm inter-plant distance within each row. Five plants within each row were selected to represent each line while the remained were discarded. The phenotyped plants within each line were considered as the subset to continue purification through single seed descent approach (SSD). The subsequent lines progeny were genotyped using a recent validated KASP-SNP assay (Cottage et al. 2013) and results will be published later separately. A set of 12 qualitative morphological traits and 20 quantitative traits were scored (Table 1). Each row was covered by a 1.3m tall insect-proof net (Provided by Diatex, France) from anthesis to the end of flowering. Hand weeding both between and withinrows was performed regularly. Pesticides: λ–Cyalothrine 25 g/L (Sygenta) and Deltamethrine 25% (Bayer Crop Science) were also applied. No fertilizers or supplementary irrigation were used. Aphid control took place daily.

1.3 Measurements
All traits (Table 2) were measured and recorded using recommended scales in accordance with the International Union for the Protection of New Varieties (UPOV, 2002) and (Biodiversity international, IBPGR, 1985) faba bean descriptors.

Table 2 Morphological traits used for the description of the field bean lines and controls

 
1.4 Statistical Analysis
Simple descriptive statistics (e.g., mean and coefficient of variation) were used to compare variation among the lines. Analysis of variance (ANOVA) was also carried out. Means were compared using least significant difference (LSD) test. All computations were carried out in Genstat 14.0 (Payne et al., 2007). Hierarchical clustering of the ninety-seven breeding lines and the three controls were based on the phenotypic traits using Euclidean distances. A dendrogram was created using Ward’s method (1964) implemented in the Xlstat (2012) software.

2 Results
2.1 Qualitative traits
There was no variation for stipule spot pigmentation; all lines and controls were spotted. All lines had low or moderately low susceptibility to lodging. Leaflet shape was mostly intermediate for 76% of lines which have SSP type 1, while 21.6% had narrow leaflets with Lpmw type 2 and only 4.1% had large leaflets with LPW type 3 on the UPOV scale (UPOV, 2002). The two controls NV813 ‘INAT-L2’ and ‘INAT-saber 02’ had intermediate leaflet shape while NV819 ‘INAT-L3’ had narrow type. Stem anthocyanin pigmentation showing two contrasting colors at the four leaf stage which were confirmed at flowering time (Figure 1). To sum up, 72% of the lines and the 3 controls varieties had green stem without pigmentation or with only slight pigmented (Table 3 and Figure 1); which correspond to scores 1, 2 and 3 on the UPOV scale. 38% of lines had intense anthocyanin pigmentation on their stem (Table 3) corresponding to UPOV scores of 5, 6; 7 and 8 (Figure 1). All the anthocyanin pigmented stems had also pigmented flower standards.

Figure 1 Stem Contrasted Color (Left: Anthocyanin Stems Pigmentation; Right: Green Stems)

Table 3 Classification of lines according to the stem anthocyanin pigmentation intensity

 
Determinate growth habit was identified in a single plant of the five scored within line 'INAT-21' (FB3-L6-P2); a single flower flags the main stem and branches apex at the end of flowering time. Furthermore, it is visible after flowering thanks to the presence of single pod at the apex instead of a vegetative node (Figure 2). All other lines and the three controls had indeterminate growth.

Figure 2 Phenotype of Determinate Growth in Line 21 ‘INAT-21’ (Left: single pod on the apex main stem apex; Right: the stem apex become a flowering node)

 
All lines and controls had wild type common flower color and flower standards with intense anthocyanin streaks (purple pigment) and a dark melanin spot on their wings. The Intensity of the flower standard streaks did not vary. Pod surface reflectance, and pod color at maturity did vary (Table 4). All controls and 54% of the lines had glossy dark pods (Figure 3) compared to 46% with a matt surface and a light color (Figure 3).

Table 4 Classification of Lines According to the Intensity of Anthocyanin

Figure 3 Pod surface reflectance and color

 
The majority of lines (96%) and the 3 controls varieties had wild type seeds with black hilum color (Figure 4). However; within the lines ‘INAT-32’, ‘INAT-31’and ‘INAT-79’ respectively two, two and one plant among 5 studied plans within each line were identified to have a white hilum (pale) seeds (Figure 4) out of the five scored for each line.

Figure 4 Hilum Color Contrast (Left: white hilum (Pale); Right: black hilum (wild type))

 
Anthocyanin spots on the seed testa (Figure 5) were observed surprisingly within two plants of the control variety ‘local population’ INAT-L2’. In order to check the stability of this trait, one spotted seed from each plant that was identified to carry this trait was sown in glass house. The resulting harvested seeds were all spotted for all the seeds of each plant (Figure 5). This trait has not been described before in the control variety NV813: ‘INAT-L2’ local which population which is supposed to be pure. Segregation for this trait has previously been unnoticed.

Figure 5 Seed Anthocyanin Pigmentation

 
Four seed testa colors were identified one month after harvesting. On average, 66.5% on average seed testa were beige, 1.2% have a yellowish-green testa. 28.9% of the seeds were light brown, and 3.4% are dark brown (Figure 6). The controls all have light brown seeds.

Figure 6 Seed Testa Color (from left to right: Yellowish-green; Beige; Light brown and dark brown)

 
The frequency distribution of seed coat colors for each genotype (Figure 7) showed that lines INAT-114-11; INAT-109-11; INAT-19-11, and INAT-17-11 are characterized by 100% beige seeds while the line INAT-88-11 differs in that 90% of the seeds are dark brown.

Figure 7 Frequency of seed coat colors within various Vicia faba var minor lines

 
2.2 Quantitative traits
Quantitative traits (Table 5) revealed highly significant variation (P < 0.01) among the nineteen quantitative descriptors. It was notable that some lines flowered 10 days earlier than the average. This could be important for spring water limiting conditions in Tunisia and other Mediterranean regions.

Table 5 Analyses of Variance (ANOVA) for Quantitative Traits

 
The final plant height (FPH) showed an overall mean of 70.80 cm (Table 4). 34% of the lines have average heights between 70 and 75 cm whereas lines that have heights between 55 and 65 cm are represented by only 20% (Figure 8a). The tallest line was 'INAT-03-11' (89.4cm) while the shortest lines were 'INAT-48-11' and 'INAT-45-11' with heights 49.8 and 48.6 cm respectively. The control cultivars: ‘INAT-L2-11’, 'INAT-SB02-11'and INAT-L3-11’ have the following heights respectively 72.8, 65.2 and 65 cm. The mean total number of nodes (TNN) on the main stem per line was 22.51(Table 5). 44% of the lines had average between 21 and 24 nodes (Figure 8b). The Line 'INAT-8-11' had the highest TNN (29 nodes) while 'INAT-96-11’ has only 11 nodes.

Figure 8 Quantitative traits distribution

 
The controls, local pop cultivar ‘INAT-L3-11’, 'INAT-SB02-11 'and‘INAT-L3-11’ had respectively 24, 18 and 22.4 nodes. Leaf Area (LA) average was 55.35 cm2/leaf (Table 5). 50% of the lines have an average between 50 and 61 cm²/leaf (figure 8c). Line 'INAT-3-11' is distinguished by the highest LA with 94.36 cm2/leaf, while 'INAT-72 had the ‘lowest LA with only 39.2 cm²/leaf. The control varieties, ‘INAT-L2-11’,'INAT-SB02-11' and ‘INAT-L3-11’ were 54, 67, 64.85 and 39.2 cm2/leaf respectively. The average main stem width (MSW) was 6.38 mm. 53% of the lines had an average between 5.70 and 6.51 mm (Figure 8d). Line 'INAT-2-11' is characterized by the largest average (10.6 mm) while 'INAT-87-11' (MSW) was only 4.2 mm. The control varieties local pop variety ‘INAT-L2-11’ ','INAT-SB02-11 'and ‘INAT-L3-11’ were 5.84, 6.4 and 5.2 mm respectively. The pod angle (PA) from the main Stem average was 30° (Figure 8e) which generally corresponds to erect pods according to the UPOV 2002 descriptors (Table 2). 35% of lines had an average between 25° and 35° (Figure 8e). The lines'-INAT: 36, 109, 55, 34, 108, 97, 23 and 71’ have the largest PA with 45°relative to the main stem, whereas, the lines 'INAT 92, 38, 39, 118, 25, 74, 86, 66, 89, 49, 103, 58, 60, 90, 44, 102, 72, 106 have the lowest PA at 20°.The control varieties ‘INAT-L2-11’,'INAT-SB02-11’ and the ‘INAT-L3-11’ have 35°, 30° and 20° respectively. We noticed that the only plant carrying determinate growth habit within the line ‘INAT-21’ had the lowest value of PA (15°) which makes pods more erect and very close to the stem. Branching (Br) from the main Stem had an overall average of 3 branches per plant (Table 4, Figure 8f). The line 'INAT-41-11' (FB3 L3P6) showed or displayed the highest number of branches per plant (5) on average. The controls ‘INAT-L3-11’, 'INAT-SB02-11 'and ‘INAT-L3-11’ had 2 branches on average. The mean Number of pods per Plant (TNP) was 17.70. The line INAT-11-11 has the highest TNP with 45 pods per plant. Controls ‘INAT-L2-11’, ‘SB02-INAT-11’ and ‘INAT-L3-11’ had respectively 17, 19 and 10 pods per plant. The majority of field beans lines (39%) have a (TNP) between 15 and 20 pods per plant. The highest performances were recorded in 1% of lines with 30~35 pods / plant, and 1% with 45~50 pods/ plant against 6% with low yield (5 to 10 pods per plant). The division of the TNP regarding the number of seeds per pod or ovules per pod: 1 (NP1S) 2 (NP2S) 3 (NP3S), and 4 (NP4S) showed that NP3S and NP4S contribute by 45% and 40% to the total pods produced per line (Figure 8g). NP1S and NP2S didn’t exceed 4 and 10% of produced pods respectively. Lines INAT-16-11; INAT-120-11; INAT-79-11; INAT and control-L2-1 had NP4S rate with respectively 78%, 76%, 65% and 66%. The majority of the lines had a 100 seed weight (100-SW) between 20 and 50 g and are therefore considered as field beans (var minor) as classified by Duc (1997). Only 1% of lines had a 100-SW between 50 and 60 g corresponding to horse beans (var equina) (Figure 8f). INAT-32-11 has the highest 100-SW (56g) vs ‘INAT-100-11’ with the lowest (27.76 g). Total Seed per plant (TSP) showed variability between lines with an overall average of 57 seeds per plant (Table 4, Figure 8f). INAT-11-11 is characterized by the highest average TSP with 149.8 seeds / plant and has the highest contribution by NP3S and NP4S with 91% of the total number of pods per plant produced. INAT-49-11; INAT-98-11; INAT-99-11; INAT-70-11, and the control INAT-L3-11 had a low TSP not exceeding 25 seeds / plant. The number of days from sowing to 50% of plants within a line with at least with one opened flower (FT) showed 12 classes (Table 6) with a maximum of 66 days for the class A and a minimum of 53 days for class L. Among the controls ‘INAT-SB02-11’ belongs to the class H with 57 days required to get 50% of plants with at least one flower; followed by INAT-L2-11 that belongs to class I. INAT-L2 belongs to class L with 53 days and is considered precocious. The difference in the FT (Table 6) was found to be as much as 13 days. In the current study, the period from sowing to flowering varied from 53 to 66 days and NDM ranged from 152 to 156 days in the superior semi-arid condition of Tunis. Early pod setting and seed development allowing escape from late season drought which is frequent in the Mediterranean region. The three white hilum lines ‘INAT-32-11,’ ‘INAT-31-11’ ’and especially ‘INAT-79-11’ showed or displayed good values of flowering precocity, with (FT) values respectively 63, 62 and 57 days compared to 53 days for the control ‘INAL-L2-11’ and 57 for a recently registered variety ‘INAT-SB02-11. Notably, the unique determinate growth plant within the line INAT-21-11 had the lowest FT at 45 days compared to the four other plants of the same line with FT of 66 days.

Table 6 NDF classes (LSD: 5%)

 
The number of days from sowing to maturity showed 5 classes (Table 7) with a maximum of 156 days for class A and a minimum of 152 days for class E. Among the controls ‘INAT-SB02-11’ belongs to class A with 156 days required for maturity followed by ‘INAT-L3-11’ in that class. INAT-L2 belongs to class D with 153 days and is considered precocious. The difference in NDM (Table 7) was as much as 4 days.

Table 7 NDM classes (LSD: 5%)

 
Significant positive correlations were observed between plant height (FPH) with the following traits: TNN (r = 0.321 **), MSW (r = 0,574**), TNP (r = 0.433 **), NP4S (0.301 **) and TSP (r = 0.407 **) respectively (Table 8).

Table 8 Correlations between traits

 
A dendogram of faba bean genotypes was created from a hierarchical classification of Euclidean distances derived from the analysis of quantitative parameters. The 4000 Euclidean distance (Figure 9), shows five groups. The first group is represented by lines INAT-X-11: (x = 15, 3, 22, 27, 8, 13, 2, 48, 24, 20 and 19) with ‘X’ refers to the numbering related to each line on the Figure 9. These faba bean lines are characterized by a similar stem height 78.10 ± 5.45 cm. Pods with 4 and 3 seeds account for 76% of pods produced by these field beans inducing a relatively high number of seeds 76.16 ± 11.3 plant.

Figure 9 The dendrogram based on Euclidean distances and Ward method

 
Group 2 is formed by the INAT-X-11 with X: 21, 10, 47, 43, 79, 83, 82, 96, 123 and 63, 93, 57, 83, 91, 26, 119, 124, 61 , 56 and 126, 81, 109, 55, 62, 92, 84, 111, 66, 120, and the control INAT-SBR02-11 and INAT-L2-11are classified together into the third group with total pods and seeds per plant, respectively, 19.95 ± 2.85 pods and 64.75 ± 9.39 seeds. These lines had in common an NP4S that represents 57% of the total produced pods. Group 3 is only formed by INAT-11-11 and is distinguished on production of pods per plant and higher seed yield with respectively 45.6 pods / plant and 149.8 seeds per plant. Group 4 clusters lines 25, 9, 16, 51, 50, 46, 17, 45 and 44, 38, 41, 40, 72, 69, 52, 121, 68, 75, 39, 29, 97, 71 , 127, 42, 23, 108, 36, and the two-line INAT 31-11 and INAT-32-11 identified as segregating for hilum color. These lines have in common a number of pods with 2 and 3 seeds more representative.

Group 5 is formed by the INAT-11-X lines with X: 115, 58, 118, 74,86, 93, 89, 60, 103, 106, 102, 90, 100, 88, 73, 114, 94, 59 , 87, 96, 49, 99, 70, 98, and 11-INAT-sbrx It is characterized by the lowest pod yield at 12.35 ± 2.55 and the lowest number of seeds per plant (40.64 ± 8.62) (Table 9).

Table 9 clustering according to traits similarity

 
3 Discussion
The lines studied are important for their potential use in field bean pre-breeding programs and as a source of diversity. The lines represent the F4 generation after selection that occurred mainly in open pollination conditions. In this case outcrossing is inevitable. Unfortunately, information regarding the pattern of genetic diversity is not available on this set of breeding lines. In order to evaluate these lines an agro-morphological characterization was carried out as the starting point for selecting improved individuals. The results of this study indicated that there is genetic variability among the 97 lines and the three controls. This variability could be utilized in a breeding program. Indeed, the lines carry some targeted qualitative traits such as determinate growth habit; characterized by a terminal inflorescence in the top of the stem. This is known to greatly decrease the number of flowering nodes after the beginning of flowering Huyghe (1998). Moreover, it has a considerable effect on reducing plant height and lodging and also promotes better partition of assimilates between vegetative and reproductive growth and increases the harvest index (Avila et al., 2007). This trait is relevant according to Nadal et al. (2005) in facilitating crop management and mechanical harvesting. Several faba bean mutations for this trait are due to a single gene (Sjodin, 1971; Filipetti, 1986; Avilla et al., 2006). A diagnostic SCAR marker developed by (Avilla et al., 2006) has been recently transformed into a SNP marker by Cottage et al. (2012). The second trait of interest; the white hilum character occurs at low frequency in the collection. Hilum colour is a simple Mendelian trait controlled by a single recessive gene with maternal control (Sirks, 1931). Faba bean consumers have a preference for white hilum. Hence, this could be a valid target for selection in its own right. Moreover, Duc et al. (2004) and Duc et al. (1989) isolated a faba bean line with 20-fold less vicine and convicine than the wild-type average. The low vicine-convicine (LVC) trait is inherited as a simple recessive Mendelian character (designated vc-) and was shown to be linked in coupling phase in the studied background (“line 1268”) to the colorless hilum character (Gutierrez et al., 2006). Thus, breeding with the vc- mutations should be straightforward, as well as both economically and socially beneficial. Some LVC cultivars have been released by INRA and are marketed by RAGT, but they are few in number and are not yet widely grown outside their home country. Other breeders have started to exploit the vc- allele in their breeding programs (Khamassi et al., 2013). Another trait identified by this screening is spotted testa (Figure 8). This trait, rarely observed in faba bean germplasm, gives another empirical indication of the great morphological diversity present in this small collection. It may also be of interest as a phenotypic trait which could be used to identify a future market segment, in much the same way as variegated testa markings define the “pinto” and “borlotto” market classes in Phaseolus bean. Further studies by genotyping and chemical analysis are needed. All the INAT collection lines had a colored flower (linked with tannin level). This result is in agreement with those of Oujiet al. (2011) and Yahia et al. (2012) who found only the colored wild type flower in Tunisian faba bean populations. The pod shape varies within some lines and between cultivars. This trait is a quality selection criterion which depends on the standards of the validation board and on consumer demand but we found no studies in which this trait is discussed for Vicia faba. All the lines are low or moderately susceptible to lodging and would therefore be acceptable for breeding because stem strength or ability to stand is important to in harvesting (Gnanasambandam et al., 2012). The stiff-stem trait is one of the targeted traits in faba beans and is controlled by a single recessive gene (Frauen and Sass, 1989). Seed coat color was mainly beige to brown as found by Ricciardi (1985) and in contrast to Duke et al. (1995) who showed the dominance of green color. Leaf area reflects the size of the leaves and has a wide variability (Lawes, 1980; Polignano et al., 1987). According to Robertson and El-Sherbeeny (1991) selection should favor lines with small to medium leaf area to maximize photosynthesis efficiency. The same author reported that breeders are in the process of collecting various resources, demonstrating the variability of this character, to select those with reduced leaf surfaces to get a higher leaf area index. Lawes (1980) reported Vicia faba L with indeterminate growth habit and which have sometimes very large values of leaf area may engender or produce shading resulting in competition for assimilates between plants thereby limiting the potential yield. The angle of the pods from the stem axis is important for the mechanization of harvesting. This should be less than 90°, preferably the pods should be as upright as possible Lawes et al. (1983). Regarding branching, the results are in agreement with Silim and Saxena (1992), Al-bariri and Shtaya (2013) and Karakoy et al. (2013) who all found wide variability of primary branching among Vicia faba. Such variation could be due to differential response to environmental conditions. Likewise, previous studies in grain legumes presented this trait as very sensitive to seed density and affected by both environmental and genetic control (Huyghe, 1998). According to Huyghe et al. (1994), a negative correlation was noticed occasionally between the main stem nutritional contribution to the yield and the branches contribution. Further studies on genotyping are needed to increase knowledge of branching pattern as this trait has nutritional effects. Some genes were discovered in some other grain legumes such as soybean (Glycine max L. Merr). Indeed, two independent genes Br1 and Br2 which act on branching were discovered (Nelson, 1996). In white lupin, the number of first–order branches is modified by the presence of a gene controlling determinate growth habit Julier and Huyghe (1993). Considering the total number of seed produced pod per plant, Lopez-Bellido et al. (2005) reported that this trait is a more important target for selection than the number of fertile nodes or the number of pods per fertile node and is generally correlated with yield. Moreover it is positively correlated with plant density (Salih, 1989; Evans, 1980; Stringi et al., 1986). Stutzel and Aufhammer (1992) reported that the number of pods per square meter was the most important factor correlated with seed yield. The large number of plants per square meter compensates for the decrease in the number of pods resulting from inter plant competition (Lopez-Bellido et al., 2005). Suso et al., (1996) reported significant differences in the number of seeds per pod among twelve genotypes of faba bean. Li-Juan et al. (1993) showed that the number of seeds per pod varies on average from 1.7 to 2.9 in a collection of 1500 accessions from different provinces of China. Abdelmola and Abuanja (2007) reported that the number of seeds per pod is influenced by genetic and environmental factors. The 100-SW is considered a major criterion for classification of botanical varieties of Vicia faba L. (Muratova, 1931; Duke, 1997). Duke (1997) reported that faba bean (Vicia faba L. var major) has a 100-SW greater than or equal to 100 g, field bean (Vicia faba L. var minor) was less than 50 g, while types with medium seed size (Vicia faba L. var equina) have a 100-SW between 50 and 100 g. This parameter varies significantly in Palestinians landraces of Vicia faba L. between 35.67 and 239 g (Al-Bariri and Shtaya, 2013). Keneni et al. (2005) found an average of 42.31 g 100-SW among Ethiopian beans. Abbes et al. (2007) showed that the 100-SW varies in faba bean Tunisian varieties from 45.93 to 68.88 g. Whereas, Karakoyet al. (2013) found large variability in this trait, ranging from 13.80 to 166.75 within mixed collection of Turkish landraces and varieties. Correlation coefficients for the agro-morphological traits revealed some associations that agree with some previous studies; Bianco et al. (1979) showed a positive correlation between the number of seeds and plant height in field beans. Ulukan et al. (2003) reported the direct and indirect effects of plant height, the number of pods per plant and number of seeds per pod on the biological yield in faba bean with a coefficient of determination (R2 =0.636). The same authors found a positive correlation between seed production and plant height on the one hand and between 100-seed weight, seed weight per plant and biological yield. Fikreselassie (2012) found a significant positive correlation between the number of seeds per plant in Vicia faba L. with plant height (r = 0.734) and number of pods per plant (r = 0.654). In this study NP3S and NP4S were found to have a highly significant contribution to the development of the number of seeds per plant with coefficients (r = 0.806) and (r= 0.718) respectively. These are so highly correlated with the TNP. Fikreselassie (2012) showed that plants which shed the greatest number of pods per plant produce more seeds. Thus, selection for the number of pods provides a significant improvement in seed yield. The dendrogram shows a phenotypic variability in perspective agronomic performance that may reflect the degry of heterozygocity due to previous outcrossing between these lines of Vicia faba L. var minor. Division of genotypes into groups (Table 8) will facilitate future work for the selection and development of new varieties. However, this first clustering will be more significant when it will be completed by the degree of purity of the same collection using SNP genotyping.

The present work constitutes a first step in describing genetic variability in the INAT collection. The identified lines with recessive qualitative trait are quite interesting, such white hilum, determinate growth habit, and spotted seed; this last trait can be useful as morphological trait during breeding exercises. We envisage in the future complementing the present study with a genotypic survey on one single selected seed progeny of this collection by genotyping with SNP (single nucleotide polymorphism) markers validated by Cottage et al. (in preparation) in order to better understand the genetic and genomic organization of this set of lines and check its purity. Consequently, this study is a first step for identify interesting lines as targets for subsequent costly analysis such as the study of the genetic linkage between while hilum and low vicine and convicne content (anti-nutritional compound in the seed) allele vc- by HPLC.

References
Aguilera-Diaz C., and Recalme-Manrique L., 1995, Effects of plant density and inorganic nitrogen fertilizer on field beans (Vicia faba), Journal of Agricultural Science, 125: 87-93
http://dx.doi.org/10.1017/S0021859600074530
 
Avila C.M., Atienza S.G., Moreno M.T., and Torres A.M., 2007, Development of a new diagnostic marker for growth habit selection in faba bean (Vicia faba L.) breeding, Theoretical and Applied Genetics 115(8): 1075-1082
http://dx.doi.org/10.1007/s00122-007-0633-y

Avila C.M., Nadal S., Moreno M.T., and Torres A.M., 2006, Development of a simple PCR-based marker for the determination of growth habit in Vicia faba. L. Using a candidate gene approach, Molecular Breeding 17:185-190
http://dx.doi.org/10.1007/s11032-005-4075-4

Benachour K., Kamel L., and Michaël T., 2007, Rôle des abeilles sauvages et domestiques (Hymenoptera: Apoidea) dans la pollinisation de la fève (Vicia faba L. var. major) (Fabaceae) en région de Constantine (Algérie). Annales de la Société Entomologique de France 43: 213-219
http://dx.doi.org/10.1080/00379271.2007.10697513

Bond D.A., 1976, Field bean, Vicia faba (Legumillosae papilionatae). In: Simmonds, N.W. (Ed.), Evolution of Crop Plants. Longman, London, UK. pp.179-182

Bond D.A., and Poulsen M.H., 1983, Pollination. In: ed. P.D. Hebblethwaite, The Faba Bean. Butterworth, London, pp. 77-101

Brown A.H.D., and Briggs J.D., 1991, Sampling strategies for genetic variation in ex-situ collections of endangered plant species, In: Falk D.A., and Holsinger K.E., (eds.), Genetics and conservation of rare plants. Oxford University Press, New York: 99-119

Cabrera A., 1988, Inheritance of flower color in Vicia faba L., FABIS Newsletter, 22: 2-7

Crofton G.R.A., and Bond D.A., 1998, A review on the genetics of seed coat colour and hilum colour in field beans (Vicia faba L.) with comments on some implications of national listing and certification, Plant Varieties And Seeds, 11: 97-100

Duc G., 1997, Faba bean (Vicia faba L.), Field Crops Research, 53: 99-109
http://dx.doi.org/10.1016/S0378-4290(97)00025-7

Duc G., Sixdenier G., Lila M., and Furstoss V., 1989, Search of genetic variability for vicine and convicine content in Vicia faba L. A first report of a gene which codes for nearly zero-vicine and zero-convicine contents. In: Recent advances of research in antinutritionnal Factors in legumes seeds, J. Huisman, A.S.M. J.F.B. Van der Poel, I.E. Liener (Eds) Pudoc, Wageningen, Netherlands (Pbs), pp.305-313

Duc G., Marget P., Page D., Domoney C., 2004, Facile breeding markers to lower contents of vicine and convicine in faba bean seeds and trypsin inhibitors in pea seeds, In: Muzquiz M., Hill G.D., Cuadrado C., Pedrosa M.M., Burbano C., (eds), Recent advances of research in antinutritional factors in legume seeds and oilseeds. Wageningen Academic Publishers, Wageningen, pp281-285

Evans L.T., 1980, The natural history of crop yield, American Scientist, 68: 388-397

Filippetti A., 1986, Inheritance of determinate growth habit induced in Vicia faba L. major by ethyl methane sulphonate (EMS), Fabis News, 15:12-14

Frauen M., and Sass O., 1989, Inheritance and performance of the stiff-strawed mutant in Vicia faba L. In Proceedings of XII EUCARPIA Congress, Göttingen, Germany, pp.13-18

Gnanasambandam A., Paull J., Torres A., Kaur S., Leonforte T., Zong X., Yang T., and Materne M., 2012, Impact of Molecular Technologies on Faba Bean (Vicia faba L.), Breeding Strategies Agronomy, 132-166

Gous R.M., 2011, Evaluation of faba beans (Vicia faba cv Fiord) as a protein source for broilers, South Africain Journal of plant Animal Science, 41(2): 71-78

Gutierrez N., Avila C.M., Moreno M.T., and Torres A.M., 2008, Development of SCAR markers linked to zt-2, one of the genes controlling absence of tannins in faba bean, Australian Journal of Agricultural Research 59: 62-68
http://dx.doi.org/10.1071/AR07019

Hebblethwaite P.D., 1983, In: Hebblethwaite P.D., (Ed.), the Faba Bean. Butterworths, London.UK. pp.573

Huyghe C., Julier B., Harzic N., and Papineau J., 1994, Yield and yield components of indeterminate autumn-sown white lupin (Lupinus albus) cv. Lunoble, European Journal of Agronomy, 3: 145-152

Huyghe C., 1993, Growth of white lupin seedlings during the rosette stage as affected by seed size, Agronomie, 13: 145-153
http://dx.doi.org/10.1051/agro:19930209

Huyghe C., Harzic N., Julier B., and Papineau J., 1994, Comparison of determinate and indeterminate autumn-sown white lupins under the western European climate, In: Dracup M., Palta J. (Eds.), Proceedings of the First Australian Lupin Technical Symposium, DAWA, Perth, Australia, pp.123-128

Huyghe C., 1998, Genetics and genetic modifications of plant architecture in grain legumes: a review, Agronomie, 18: 383-411
http://dx.doi.org/10.1051/agro:19980505

Khamassi K., Ben Jeddi F., Hobbs D., Irigoyen J., Stoddard F., O’Sullivan D.M., and Jones H., 2013, A baseline study of vicine-convicine levels in faba bean (Vicia faba L.) germplasm, Plant Genetic Resources News Archive, 11(03): 250-257
http://dx.doi.org/10.1017/S1479262113000105

Keneni G., Mussa J., Tezera W., and Getnet D., 2005, Extent and pattern of genetic diversity for morpho-agronomic traits in Ethiopian highland pulse landraces II. Faba bean (Vicia faba L.). Genetic Resources and Crop Evolution, 52: 551-561
http://dx.doi.org/10.1007/s10722-003-6022-8

Link W., Dixkens C., Singh M., and Schwall M., 1995, Genetic diversity in European and Mediterranean faba bean germplasm revealed by RAPD markers, Theoretical and Applied Genetics, 90: 27-32
http://dx.doi.org/10.1007/BF00220992

Lopez-Bellido F.J., Lopez-Bellido L., Lopez-Bellido R.J., 2005, Competition, growth and yield of faba bean. European Journal of Agronomy, the Journal of the European Society for Agronomy, 23(4): 359-378

Martini A., Migliorini P., Lorenzini G., Lotti C., Rosi Bellière S., Squilloni S., Riccio F., Giorgetti A., and Casini M., 2008, Production of grain legume crops alternative to soya bean and their use in organic dairy production. In the 6th IFOAM Organic World Congress, Modena, Italy

Nadal S., Cabello A., Flores F., and Moreno M.T., 2005, Effect of growth habit on agronomic characters in faba bean, Agriculturae Conspectus Scientificus, 70(2): 43-47

Naguib N.A., 2000, Morphological and chemical identification of new varieties of some field crops, Ph. D. Thesis, Faculty of Agric. Ain Shams University

Nelson R., 1996, The inheritance of a branching type in soybean, Crop Science 36: 1150-1152
http://dx.doi.org/10.2135/cropsci1996.0011183X003600050014x

Ouji A., Rouaissi M., Raoudha A., and E.L. Gazzah M., 2010, The use of reproductive vigor descriptors in studying genetic variability in nine Tunisian faba beans, African Journal of Biotechnology, 10: 896-904

Pilbeam C.J., Aktase J.K., Hebblethwaite P.D., and Wright S.D., 1992, Yield production in two contrasting form of spring-sown faba bean in relation to water supply, Field Crops Research, 29: 273-287
http://dx.doi.org/10.1016/0378-4290(92)90030-D

Polignano G.B., Spagnoletti Zeuli P.L., 1985, Variation and covariation in Vicia faba L. populations of Mediterranean origins, Euphytica, 34: 659-668
http://dx.doi.org/10.1007/BF00035401

Payne R.W., D.A. Murray, S.A. Harding, D.B. Baird, and Soutar D.M., 2007, Genstat for Windows (10th Edition) Introduction. VSN International, Hemel Hempstead

Ramsay G.R., and Pickersgill B., 1986, Interspecific hybridisation between Vicia faba and other species of Vicia: approaches delaying embryo abortion, Biol. Zentralbl 105: 171-179

Robertson L.D., and EL-Sherbeeny M., 1991, Distribution of discreetly scored descriptors in a pure line faba bean (Vicia faba L.) germplasm collection, Euphytica, 57: 83-92

Ruggiero C., Stefania D.P., and Massimo F., 1999, Plant and soi*l resistance to water flow in faba bean (Vicia faba L. major Harz.). Plant Soil, 210: 219-231
http://dx.doi.org/10.1023/A:1004690101953

Salih F.A., 1989, Effect of sowing date and plant population per hill on faba bean (Vicia faba) yield, FABIS Newsletter, 23: 15-19

Silim S.N., and Saxena M.C., 1992, Comparative performance of some faba bean (Vicia faba L.) cultivars of contrasting plant types. 2. Growth and development in relation to yield. Journal of Agricultural Science 118: 333-342
http://dx.doi.org/10.1017/S0021859600070702

Sirks M.J., 1931, Beitrage zu einer genotypischen analyse der ackerbohne Vicia faba L., Genetika, 13: 210-631

Stützel H., and Aufhammer W., 1992, Grain yield in determinate and indeterminate cultivars of Vicia faba with different plant distribution patterns and population densities, The Journal of Agricultural Science, 118: 343-352
http://dx.doi.org/10.1017/S0021859600070714

Stringi L., Sarno R., Amato G., and Cristina L., 1986, Effects of plant density on Vicia faba L. eqzcina and Vicia faba L. minor in a semiarid environment in Southern Italy, FABIS Newsletter, 15: 42-45

Suso M.J., Moreno M.T., and Cubero J.I., 1993, New isozyme markers in Vicia faba: inheritance and linkage. Plant Breeding 40: 105-111

Torres A.M., Avila C.M., Gutierrez N., Palomino C., Moreno M.T., and Cubero J.I., 2010, Marker-assisted selection in faba bean (Vicia faba L.). Field Crops Research, 115:243-252
http://dx.doi.org/10.1016/j.fcr.2008.12.002

Tufarelli V., Khan R.U., and Laudadio V., 2012, Evaluating the suitability of field beans as a substitute for soybean meal in early-lactating dairy cow: production and metabolic responses, Journal of Animal Science, 83(2): 136-40
http://dx.doi.org/10.1111/j.1740-0929.2011.00934.x

Terzopoulos P.J., Kaltsikes P.J., and Bebeli P.J., 2003, Collection, evaluation and classification of Greek population of faba bean (Vicia faba L.), Genetic Resources and Crop Evolution, 50: 373-381
http://dx.doi.org/10.1023/A:1023962618319

Terzopoulos P.J., and Bebeli P.J., 2008, Genetic diversity of Mediterranean faba bean (Vicia faba L.) with ISSR markers, Field Crops Research, 108: 39-44
http://dx.doi.org/10.1016/j.fcr.2008.02.015

Terzopoulos P.J., Kaltsikes P.J., and Bebeli P.J., 2004, Characterization of Greek populations of faba bean (Vicia faba L.) and their evaluation using a new parameter, Genetic Resources and Crop Evolution, 51: 655-662
http://dx.doi.org/10.1023/B:GRES.0000024654.89373.c2

Yahia Y., Guetat A., Walid E., Ferchichi A., Yahia H., and Loumerem M., 2012, Analysis of agromorphological diversity of southern Tunisia faba bean (Vicia faba L.) germplasm, African Journal of Biotechnology 11(56): 11913-11924

Witcombe J.R., 1981, Genetic resources of faba beans. In: Hawtin G. and Webb C. (eds), Faba Bean Improvement. ICARDA, Aleppo, Syria, pp. 1-13

Zeid M., Schon C.C., and Link W., 2003, Genetic diversity in recent elite faba bean lines using AFLP markers, Theoretical and Applied Genetics, 107: 1304-1314
http://dx.doi.org/10.1007/s00122-003-1350-9