Genetic Resources and Crop Evolution 50: 373–381, 2003.  2003 Kluwer Academic Publishers. Printed in the Netherlands.

373

Collection, evaluation and classification of Greek populations of faba bean (Vicia faba L.) P.J. Terzopoulos, P.J. Kaltsikes* and P.J. Bebeli Department of Plant Breeding and Biometry, Agricultural University of Athens, Iera Odos 75, Athens 11855, Greece; * Author for correspondence (e-mail: kaltsikes@ aua.gr; phone: 130 10 5294621; fax: 5294622) Received 7 June 2001; accepted in revised form 25 January 2002

Key words: Cross-pollinated species, Dissimilarity coefficients, Genetic resources, Germplasm, Multivariate analysis, Vicia faba L.

Abstract Fifty-five Greek Vicia faba L. populations, collected from diverse areas, were planted at two dry and low fertility sites for evaluation and classification. Yield evaluation, which was carried out by Principal Component Analysis (PCA) on the basis of seven yield traits, showed the number of pods per plant, number of ovules and seeds per pod, and branching from the basal nodes to be the most important traits for population evaluation regarding yield. For population classification, four dissimilarity coefficients (Manhattan, Average Taxonomic Distance, Euclidean distance and squared Euclidean distance) and four multivariate methods (PCA, UPGMA, Neighbor-joining and Principal Coordinate Analysis) were evaluated using fifteen morphological and seven yield traits. Neighborjoining was chosen as the most suitable multivariate method. This method combined with PCA for the seven yield traits, placed the populations into six groups. As revealed by the application of PCA on all twenty-two traits the grouping was based mainly on pod characteristics, stem thickness, plant height, 1000 seed weight and branching from basal nodes. Based on the results of the present study, a model is proposed for conserving cross-pollinated species, such as faba bean.

Introduction Faba bean (Vicia faba L.) is a grain legume cultivated for multiple usage because of its high nutritional value and its ability to grow over a wide range of climatic and soil conditions (Bond et al. 1980; Lawes et al. 1983). It performs relatively well under low rainfall conditions and it is thus competitive to wheat (Lazarou and Roupakias 2000). To date there is no known ancestor of faba bean and it cannot be crossed with other related species (Link et al. 1995; Witcombe 1981). As a result faba bean breeding is based only on populations within the species. Due to the continuous increase of genetic erosion it is necessary to collect, describe and efficiently handle the local populations on which the breeding of cultivated plants will be based (Burton 1979). There is, therefore, need to

describe the genetic variation of such populations and also to maintain them in situ and ex situ. In the past, surveys were conducted to describe the genetic variation and the handling of local collections of faba bean in Ethiopia (Mulat 1998), Egypt (Bakheit and Mahdy 1998) and Soudan (Khashmelmous 1989). Also Sindhu (1985) described the genetic variation of varieties from all over the world while Polignano et al. (1999) described the extent and patterns of the phenotypic diversity existing in a large sample of the Bari, Italy, faba bean collection. In the case of the ex situ conservation, problems such as duplicate accessions, reduction of total genetic variation, classification and conservation costs, require solutions. It is therefore necessary to devise efficient, reliable and inexpensive handling models in order to solve these problems. In establishing such a model,

374 researchers are faced with the question of which of the many existing proximity coefficients and multivariate methods to use. Sneath and Sokal (1973) underline the difficulty in choosing the best coefficient: ‘‘The choice of the coefficient will frequently be guided by the scale of the data matrix for which a pairwise similarity function must be computed’’. Concerning the choice of the most suitable multivariate method ‘‘the optimality criterion would guide our selection of the type of classifactory algorithm to be employed’’ and ‘‘no general agreement on the optimal classification exists except in cladistics’’ (Sneath and Sokal 1973). Generally, most researchers have used, when selecting morphological and yield traits, as dissimilarity coefficients, the Average Taxonomic (Tatineni et al. 1996), Euclidean (Bisht et al. 1998) Standard taxonomic, Goodman and Mahalanobis distance (Beer et al. 1993) and also the Gower coefficient (Franco et al. 1997; Johns et al. 1997). As for multivariate methods, Franco et al. (1997) reviewed and examined the performance of different clustering analyses. According to them, Ward’s method is the best strategy when the sizes of the groups are similar and UPGMA is appropriate when the groups are of different sizes. Other researchers have used as multivariate methods UPGMA (Huaman et al. 1999; Koutsos and Koutsika-Sotiriou 2001; Tatineni et al. 1996) and Principal Component Analysis (Grenier et al. 2001; Ortiz et al. 1998). The Neighbor-joining method and Principal Coordinate Analysis are more often applied when dealing with a two-stage type of data. Faba bean has been cultivated in Greece since antiquity (Roupakias 1983). Even though there are many local populations, no description of the variation of this species in Greece is available. Furthermore, faba bean could be used for food and feed in many marginal areas which have been abandoned by the farmers because of low yields. For these areas it would be useful to find populations which yield well under dry and low soil fertility conditions, either for outright growing or improvement through breeding. To do this, populations should be classified by giving more weight to yield. The aims of the present study were: 1) the collection of local Greek populations of faba bean, 2) the evaluation of these populations and the identification of the traits which could form the basis for future breeding plans under dry and low soil fertility conditions and 3) the classification of the populations considering a) the peculiarities of Greek

faba bean populations and b) giving more weight to yield traits. To approach the best classification of these populations, different dissimilarity coefficients, multivariate methods and their combinations were evaluated.

Materials and methods Plant material This consisted of fifty-five local faba bean (Vicia faba L.) populations and the commercial variety ‘Aguadulce’ as the control. Twenty-four of the local populations were obtained from the Hellenic Gene Bank of Thessaloniki, while the remainder were collected from different locations in Greece by P.J. Terzopoulos and are maintained at the Plant Breeding and Biometry Department of the Agricultural University of Athens (Table 1). These populations were cultivated during the 1998-1999 season in the experimental field of the Agricultural University of Athens and the seeds harvested were used for the experiments described below. Experimental data These were obtained from Randomized Complete Block experiments established at two sites (Agii Anargiri and Marousi) in Attica, Greece. Each experiment consisted of three blocks. In each block four plants from each population were planted. Thus the total number of plants studied per population was twenty-four. The distances from plant to plant and from row to row were 30 cm. The date of sowing at Agii Anargiri was December 5, 1999 and December 12, 1999 at Marousi (at this site the plants were placed in isolation cages). No fertilizers or any other soil additives were applied. Measurements Observations on twenty-two characteristics were taken, based on the Faba bean Descriptors (IBGR / ICARDA. 1985). Twelve of the measurements were continuous while the remaining ten were ordinate (Table 2). Z-transformation was performed on the averages of the values of all traits to meet the requirements of independence and normal distribution with zero mean (Sneath and Sokal 1973). Observations

375 Table 1. Geographical data and 1000 seed weight (per population) of the faba bean populations studied (the values in parenthesis are the accessions codes of the Gene Bank). Collection Code

Altitude (m)

Latitude (E)

Longitude (N)

1000 seedweight (g)

Collection Code

Altitude (m)

Latitude (E)

Longitude (N)

1000 seedweight (g)

0 1 2† 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20† 21 22 23 24 27 28 29

30 30 20 20 430 100 180 380 140 950 560 520 680 680 220 820 40 370 10 10 10 37 700 850 500 850 420 800

358059 358059 358129 358129 358149 388549 388549 388499 388529 398219 398419 378479 378339 378339 378399 378459 378389 378129 378159 378189 378189 398089 398029 398029 398089 398029 398009 398099

258159 258159 278369 278369 248659 228459 218309 218309 218409 218409 258159 228359 228359 228359 228109 228209 228259 238159 278209 278209 238359 218409 218459 218409 218459 218459 218459 218459

1180.12 1259.17 700.75 1234.77 1117.28 1176.22 1366.15 1132.45 1077.54 1212.21 1350.35 1315.62 1475.05 1522.41 1301.80 1070.34 1168.63 1636.99 1117.66 1416.39 630.38 1120.77 1188.13 1358.59 1022.09 1221.50 1281.87 987.34

30 31 32 (4120) 33 (4121) 34 (4122) 35 (4123) 36 (4124) 37 (4126) 38 (4128) 39 (4130) 40 (4133) 41 (4134)† 42 (4135)† 43 (4136) 44 (4137) 46 (4139) 47 (4140) 48 (4141) 49 (4142) 50 (4143) 51 (4144) 52 (4148) 53 (4149) 54 (4150) 58 (2379) 59 (2380) 60† 99‡

420 780 200 240 520 180 200 320 220 340 200 40 40 310 300 220 600 60 40 380 30 260 580 300 200 60 830 900

398009 398049 398069 388389 388429 388579 388599 388179 388199 388319 388329 388029 388029 388509 388549 388189 388549 388339 388329 388579 388459 388059 388289 388409 388309 408549 388459 398119

218459 218409 218459 238009 228309 238009 228109 238459 238459 238009 238359 248459 248459 238509 238559 248459 248009 248209 248209 238559 238359 248359 228309 228459 228409 238009 238159 218459

1159.88 989.86 1186.42 1243.18 1572.96 1571.55 1383.57 1134.32 1046.44 1199.62 1423.49 867.41 714.19 1279.84 1319.60 1092.49 1059.93 1114.36 1095.49 1079.82 1556.48 1197.52 946.12 1104.75 1000.57 1107.99 832.03 1357.65

† minor populations ‡ Commercial variety ‘Aguadul’ (‘Super Aguadulce’ from ‘Golden West’ seed company)

were also recorded on flower, wing petal and hilum color. These three characteristics were monomorphic for all populations (the color of flower being white, the wing petals spotted and the hilum color black).

Yield evaluation In order to find the most important yield traits under dry and low soil fertility conditions, Principal Component Analysis (PCA) (Sneath and Sokal 1973) was used on the following seven yield traits: a) six, which are sensitive to soil moisture (Hebblethwaite 1982; Hebblethwaite et al. 1984) namely, flowers per inflorence, pods per nod, ovules per pod, seeds per pod, pods per plant, 1000 seed weight and b) branching from basal nodes which correlates significantly with the other six traits (Bakheit and Mahdy 1998).

Dissimilarity coefficient Four dissimilarity coefficients were used for the genetic distance between the populations, namely Manhattan, Average Taxonomic distance, Euclidean distance and squared Euclidean distance (Sneath and Sokal 1973). The standardized averages of the values of every characteristic for each population were used to construct the matrix from which these coefficients were calculated. Discrimination method To classify the populations, it was decided to use the combination of the Principal Component Analysis results for the seven yield traits (thus giving more weight to yield) and the results of the most suitable multivariate method for all twenty-two characteristics

376 Table 2. The twenty-two plant characteristics utilized for the description of the faba bean populations.

for population discrimination PCA was applied on all 22 characteristics.

Continuous characteristics Branching from basal nodes Plant height (cm) Height of the lower pod (cm) Pods per nod Pods per plant Leaflets per leaf Stem thickness (cm) Ovules per pod Seeds per pod 1000 seed weight (g) Flowers per inflorescence Pod length

Results Range of measurements The range of the measurements, their averages and variances for the continuous characteristics are shown in Table 3.

Ordinate characteristics

Scale

Leaflet shape Stem colour at maturity Intensity of streaks on standard petal Pod angle at maturity Pod shape Pod colour at maturity Ground color of testa Stem pigmentation at flowering time Resistance to lodging Pod distribution on the stem

1, 2, 3 1, 2 0, 3, 5, 7 1, 2, 3 1, 2, 3 1, 2 2, 3, 4, 5 0, 3, 5, 7 3, 5, 7 1, 2, 3

of Table 2. Four methods were tested in order to find the most appropriate multivariate method, namely Principal Component Analysis, UPGMA (Sneath and Sokal 1973), Neighbor-joining (Saitou and Nei 1987) and Principal Coordinate Analysis (Sneath and Sokal 1973) using the statistical software NTSYS-pc (Rohlf 1998) and JMP (SAS Institute Inc, Cary, NC, USA 1996). In order to identify the most important traits

Yield evaluation Since the aim of the present study was to evaluate the populations regarding their yield and its components, seven characteristics which correlate directly with productivity (total weight of seeds per m 2 ) were chosen and utilized in the Principal Component Analysis. The results (Figure 1) show that the populations, according PCA-1, separated into two large groups: A and B. Based on the high coefficients on the PCA-1 (according to Brown (1991)) a coefficient can be generally considered high if its value is higher than 0.3) the characteristics which played a major role in grouping the populations were the number of branches from the basal nodes, number of ovules and seeds per pod and number of pods per plant (Table 4). Criterion of the choice of the discrimination method Faba bean populations have been traditionally divided

Table 3. Range (maximum and minimum values), averages and variance of the continuous characteristics of the fifty-five faba bean populations and variety ‘Aguadulce’, on a per plant basis. Traits

Max

Min

Average

Variance

Flowers per inflorescence Leaflets per leaf Plant height (cm) Stem thickness (cm) Height of the lower pod (cm) Pods per nod Branching from basal nodes Pod length (cm) Ovules per pod Pods per plant Average seed weight per plant (g) Seeds per pod

7 8 75 5 24 2 12 21 8 7 35.62 13

1 4 16 0.5 2 1 1 3 1 1 0.47 0

2.68 5.49 39.04 2.17 9.44 1.07 3.92 9.19 3.52 4.16 9.34 2.14

0.78 0.55 79.31 0.37 9.81 0.07 3.34 7.41 1.71 12.24 37.737 1.22

377

Figure 1. The Principal Components Analysis based on seven characteristics that correlate directly with yield of fifty-five faba bean populations and the variety ‘Aguadulce’.

into Vicia faba minor, Vicia faba equina and Vicia faba major on the basis of 1000 seed weight. Based on this, according to the limits proposed by Henelt (cited by Lawes et al. (1983)), five populations (2, 20, 41, 42, 60) fall into the Vicia faba minor category (Table 1). It was decided, therefore, that the discrimination method to be used for classifying the collected faba bean populations should at least keep these populations into one group. Population grouping using all characteristics In order to choose the most suitable discrimination Table 4. The Principal Component coefficients (PC-1) based on seven yield traits of fifty-five faba bean populations and the variety Aguadulce. Traits

PC-1

Flowers per inflorescence Pods per nod Branching from basal nodes Ovules per pod Seeds per pod Pods per plant 1000 seed weight

0.09 0.05 0.45 20.58 20.50 0.42 0.20

method, Principal Coordinate Analysis (PCOORDA), UPGMA and Neighbor-joining were used employing four dissimilarity coefficients, each one in turn. Each of the above methods produced essentially similar results for all coefficients (data not shown). The results of the application of the UPGMA, Neighborjoining, PCOORDA and PCA on all twenty-two characteristics showed that the Neighbor-joining method offered the best segregation since it placed all the minor populations in the same group, group IV (Figure 2). For this reason Neighbor-joining was deemed as the most suitable discrimination method. When this method was used, the fifty-five populations were separated into four groups, I, II, III and IV.

Population grouping giving more weight to yield Based on the combination of the results from the Neighbor-joining method for all characteristics (Figure 2) and the PCA for the seven yield traits (Figure 1), the populations could be placed into five groups while population 2 stands on its own in a separate group (Table 5). The application of PCA on all twenty-two characteristics placed the majority of

378 (Figure 3). Since this grouping is essentially the same as that of Table 5, the coefficients of the various traits with values greater than 0.3 (Table 6) provide an indication as to which traits play a major role in the placement of the populations in the six groups of Table 5. Along the PCA-1 these are stem thickness and pod length, the number of ovules per pod and pod angle at maturity while for the PCA-2 the corresponding traits are plant height, height of the lower pod, branching from basal nodes and the 1000 seed weight.

Discussion Greece has many mountainous and marginal areas where the high yielding varieties of the plants used in modern agriculture cannot be used, due to low soil fertility and lack of irrigation water. For such areas it is imperative to find cultivated plants capable of producing acceptable yields without additional inputs. The faba bean is such plant. It can be used for food and feed and since it fixes atmospheric nitrogen it does not require additional nitrogen fertilization. To produce varieties of this plant, which perform well under the conditions prevailing in the above mentioned areas, locally adapted populations should be the starting point. The collection of such populations is also necessary due to the genetic erosion that has been taking place in the past years, as the areas in which faba beans were grown in Greece have been lost to agriculture. In Greece, in spite of the fact that the faba bean has been cultivated since antiquity, it is not easy to find local populations all over the country. This is probably due to the fact that favism created health problems for some local human populations who did not choose to cultivate and consume faba beans. During the explo-

Figure 2. Dendrogram on the basis of Neighbor-joining for twentytwo characteristics in fifty-five populations of faba bean and the variety ‘Aguadulce’.

populations of group IA (Table 5) together to the right of the PCA-1 (Figure 3, areas 2 and 4). For group IB, the majority of the populations fall together in areas 1 and 2 (Figure 3). The populations of group II were placed in area 1. Finally all populations of groups III, IVA and IVB, (except population 58) can be found in area 3. Groups IA, IB, II, III, IVA and IVB, seen in Table 5, with the exception of some populations of groups IA and IB and population 58 could, therefore, be separated on the basis of PCA-1 and PCA-2

Table 5. The final assignment of the fifty-five faba bean populations and the variety ‘Aguadulce’ to the six groups on the basis of the combination of the Neighbor-joining method, based on the twenty-two morphological traits, and the PCA based on seven yield traits. Group

Populations

IA IB II III IVA IVB

0 6 3 24 2 20

1 8 9 31

4 12 11 58

5 13 22

41

42

60

7 14 27

10 17 30

15 19 35

16 23

18 28

21 34

29 36

32 38

33 43

37 44

39 46

40 48

47 52

49

50

51

53

54

59

99

379

Figure 3. The Principal Components Analysis based on the twenty-two characteristics in fifty-five populations and the variety ‘Aguadulce’.

Table 6. The Principal Component coefficients (PC-1 and PC-2) based on the entire set of the morphological traits. Traits

PC-1

PC-2

Flowers per inflorescence Leaflets per leaf Plant height Stem thickness Height of the lower pod Pods per nod Branchng from basal nodes Pod length Ovules per pod Seeds per pod Pods per plant 1000 seed weight Leaflet shape Resistance to lodging Pod angle at maturity Pod distribution on the stem Stem colour at maturity Pod colour at maturity Pod shape Ground color of testa Stem pigmentation at flowering time Intensity of streaks on standard petal

0.00028 0.07 0.11 0.30 0.16 20.22 20.08 0.39 0.32 0.25 20.29 0.14 0.11 20.07 0.31 0.27 0.25 0.02 0.29 0.10 20.11 0.12

0.18 0.26 0.32 0.09 0.34 20.20 0.37 20.17 20.28 20.20 0.11 0.37 20.16 0.19 20.08 0.12 20.02 0.03 0.19 0.22 0.15 0.004

ration phase of this project, large areas of the country were found which were devoid of local populations of faba bean. In the areas where favism does not cause problems to humans, local faba bean populations with high variability were found. The collection, characterization and maintenance of large numbers of local populations require a wellorchestrated effort which must be financially viable. This usually necessitates the assignment of the populations into few groups and the use of small numbers of plants per group for their propagation and maintenance. This must be accomplished without significant loss of genetic variability. The use of small numbers of plants is also necessary because of the many rare local populations. In many such cases only very few seeds with low germination ability were available. During this project, the simultaneous evaluation and classification of accessions with different numbers of seeds, posed an additional problem. The number of twenty-four plants per population used in this study appears satisfactory, since according to Yonesawa (cited by Brown and Briggs 1991) even ten plants can be assumed to constitute a representative sample. To discriminate among the collected populations, it was necessary to choose the most suitable dissimilari-

380 ty coefficient and multivariate method based on a realistic criterion. Therefore the 1000 seed weight was chosen, as it has been used for years in separating the subgenera minor from the other subgenera of faba beans. Other criteria (like the color of the seeds) have been proposed and utilized but the 1000 seed weight continues to be in use (Lawes et al. 1983). In the present study it was decided to give more weight to yield for grouping the populations. This was accomplished by a combination of the Neighbor-joining method on all characteristics and the PCA on seven yield traits. Among the characteristics that correlate directly with yield, the number of pods per plant, the number of ovules and seeds per pod and the number of branches per plant were the most important for the segregation of these populations under dry conditions and low fertility soils. A suitable grouping of the populations is necessary in order to create core collections which have been proposed as a means for increasing the efficiency of utilization and management of germplasm collections (Liu et al. 1999). Generally there are two types of collections: international and national. Core collections are not entirely satisfactory for national collections (like in Greece), even for large ones. The handling of the collections in these cases should be linked more closely with the in situ conservation than for international collections. The following steps are suggested for population handling in cross-pollinated species like faba bean: 1. Classify the accessions into a number of groups; this number should be significantly lower than the initial number of the accessions. The classification should be based on a well-defined criterion. In the present study the accessions were classified giving more weight to yield components under dry and low soil fertility conditions. The classification is important in order to avoid duplicate accessions or very closely related accessions concerning the characteristics, which are more important for the researcher. The classification using morphological or yield data is necessary because it is not appropriate to group the populations, every time, according to geographical data. In countries like Greece for example, one can have similar conditions prevailing in widely seperated areas while adjacent areas may differ in their climatic and soil conditions. 2. Mix equal numbers of seeds from each original collected population belonging to the same group. This number should be around 50 so that one copy

of each allele that occurred in the population with a frequency greater than 0.05 can be expected to be included in the mixture with a 0.95 level of probability (Lawrence et al. 1995). As a result a new ‘‘population’’ is created in which, theoretically the majority of the alleles of the original population belonging to this group are present. 3. If the initial seed supply of each new ‘‘population’’ is low, have some farmers (or research centers) multiply the seed in the first year and then distribute it. 4. Grow all new ‘‘populations’’ in different sites. This will probably result, after some years, in changes in allele frequencies as a result of natural selection. These new ‘‘population’’ can be collected and classified together with the other accessions of the gene bank. 5. Continue monitoring these ‘‘population’’. These new ‘‘populations’’ could be viewed as a kind of a ‘‘core collection’’ which is not maintained in the gene bank but in the field. The advantages of this scheme for the breeders are that they can use accessions, which are in the gene bank but also these ‘‘populations’’ which are in the field. The new ‘‘populations’’ can also be used in mass selection breeding schemes. Farmers can grow these locally adapted ‘‘populations’’ with reduced inputs and perhaps organically. It is of course necessary to maintain the initial populations as a reserve collection (as Brown 1995 proposed creating a core collection) so that all initial variation is preserved.

Acknowledgements We thank N. Stavropoulos and S. Samaras for supplying seeds and providing useful information for the landraces used in this study.

References Bakheit B.R. and Mahdy E.E. 1998. Variation, correlations, and path-coefficient analysis for some characters in collections of faba bean (Vicia faba L.). FABIS Newsletter 20: 9–14. Beer S.C., Goffreda J., Phillips T.D., Murphy J.P. and Sorrels M.E. 1993. Assessment of genetic variation in Avena sterilis using morphological traits, isozymes, and RFLPs. Crop Sci. 33: 1386– 1393. Bisht I.S., Mahajan R.K., Loknathan T.R. and Patel D.P. 1998. Diversity in Indian sesame collection and stratification of germ-

381 plasm accessions in different diversity groups. Genet. Resour. and Crop Evol. 45: 325–335. Bond D.A., Lawes D.A. and Poulsen M.H. 1980. Broadbean (Faba bean). In: Fehr W.R. and Hadley H.H. (eds), Hybridization of Crop Plants. American Society of Agronomy and Crop Science Society of America, Madison, Winconsin, USA, pp. 203–213. 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, Oxford, pp. 99–119. Brown A.H.D. 1995. The core collection at the crossroads. In: Hodgkin T., Brown A.H.D., van Hintum Th.J.L. and Morales E.A.V. (eds), Core Collections of plant Genetic Resources. IPGRI / Wiley-Sayce Publication, United Kingdom, pp. 3–19. Brown J.S. 1991. Principal Components and Cluster Analysis of cotton variability across the U.S. Cotton Belt. Crop Sci. 31: 915–922. Burton G.W. 1979. Handling cross-pollinated germplasm efficiently. Crop Sci. 19: 685–690. IBGR / ICARDA. 1985. Faba Bean Descriptors, Rome, Italy. Franco J., Crossa J., Villasenor J., Taba S. and Eberhart S.A. 1997. Classifying Mexican Maize accessions using hierarchical and density search methods. Crop Sci. 37: 972–980. Grenier C., Bramel-Cox P.J. and Hamon P. 2001. Core Collection of sorghum: I. Stratification based on eco-geographical data. Crop Sci. 41: 234–240. Hebblethwaite P.D. 1982. The effects of water stress on the growth, development and yield of Vicia faba L. In: Hawtin G. and Webb C. (eds), Faba Bean Improvement. ICARDA, Aleppo, Syria, pp. 165–175. Hebblethwaite P.D., Scott R.K. and Kogbe J.O.S. 1984. The effect of irrigation and bees on the yield and yield components of Vicia faba L. In: Hebblethwaite P.D., Dawkins T.C.K., Health M.C. and Lockwood G. (eds), Vicia faba: Agronomy, Physiology and Breeding. Martinus Njhoff / Dr W. Junk Publishers, Hague, Netherlands, pp. 71–93. Huaman Z., Aguilar C. and Ortiz R. 1999. Selecting a Peruvian sweetpotato core collection on the basis of morphological, ecogeographical, and disease and pest reaction data. Theor. Appl. Genet. 98: 840–844. Johns M.A., Skroch P.W., Nienhuis J., Hinrichsen P., Bascur G. and Munoz-Schick C. 1997. Gene pool classification of common bean landraces from Chile based on RAPD and morphological data. Crop Sci. 37: 605–613. Khashmelmous A.E. 1989. Evaluation of faba bean (Vicia faba) varieties in the heavy clay soils of Central Sudan (Sennar Area). FABIS Newsletter 23: 10–12. Koutsos T.V. and Koutsika-Sotiriou M. 2001. Genetic diversity in

four cabbage populations based on UPOV and IPGRI description forms and allozyme variation. Journal of Agricultural Science, Cambridge 136: 309–318. Lawes D.A., Bond D.A. and Poulsen M.H. 1983. Classification, origin, breeding methods and objectives. In: Hebblethwaite P.D. (ed.), Faba Bean. Butterworth-Heineman, pp. 23–76. Lawrence M.J., Marshall D.F. and Davies P. 1995. Genetics of genetic conservation. I. Sample size when collecting germplasm. Euphytica 84: 89–99. Lazarou E. and Roupakias D. 2000. Effect of plant density (number of plants per 1000m 2 ) and the between row distance on grain yield of two faba bean varieties (Vicia faba L.). Agricultural Research 23 In Greek with abstract in English: 7–14. Link W., Dixkens C., Singh M., Schwall M. and Melchinger A.E. 1995. Genetic diversity in European and Mediterranean faba bean germplasm by RAPD markers. Theor. Appl. Genet. 90: 27–32. Liu F., von Bothmer R. and Salomon B. 1999. Genetic diversity among East Asian accessions of the barley core collection as revealed by six isozyme loci. Theor. Appl. Genet. 98: 1226– 1233. Mulat G. 1998. Variation among Ethiopian faba bean landraces for seed yield and agromorphological characters. FABIS Newsletter 41: 05–08. Ortiz R., Ruiz-Tapia E.N. and Mujica-Sanchez A. 1998. Sampling strategy for a core collection of Peruvian quinoo germplasm. Theor. Appl. Genet. 96: 475–483. Polignano G.B., Alba E., Uggenti P. and Scippa G. 1999. Geographical patterns of variation in Bari faba bean germplasm collection. Genet. Resour. and Crop Evol. 46: 183–192. Rohlf F.J. 1998. NTSYS-Numerical Taxonomy and Multivariate Analysis System. Exeter Publ., New York. Roupakias D.G. 1983. Faba beans in Greece: past and future. FABIS Newsletter 7: 6–8. Saitou N. and Nei M. 1987. The Neighbor-joining Method: A new method for reconstructing phylogenetic Trees. Mol. Biol. Evol. 04: 406–425. SAS Institute Inc, Cary, NC, USA 1996. JMP/ Sales Department. Sindhu J.S. 1985. Multivariate analysis in faba bean (Vicia faba L.). FABIS Newsletter 12: 5–7. Sneath P.H.A. and Sokal R.R. 1973. Numerical Taxonomy. W.H. Freeman and Company, San Francisco. Tatineni V., Cantrell R.G. and Davis D.D. 1996. Genetic diversity in elite cotton germplasm determined by morphological characteristics and RAPDs. Crop Sci. 36: 186–192. 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.