Interrelationships among traits and morphological diversity of wheat ( Triticum aestivum L . ) accessions in base collection of Plant Genetic Resources Institute , Albania

The object of the study was the evaluation of the morphological variability of 92 wheat germplasm (Triticum aestivum L.) part of ex situ collection of Plant Genetic Resources Institute, Agricultural University of Tirana. Principal components and cluster analysis were carried out involving 8 quantitative traits, such as tiller capacity, plant height, spike length, number of spikelet per spike, number of seeds per spikelet, number of seeds per spike, seed size and of seeds per spike. Plant height showed positive significant correlation with yield contributing traits as spike length (r = 0.560) and the number of spikelet’s per spike (r = 0.305). The number of grains per spike had a significant positive relationship with the mass of grains per spike. Three principal components exhibited about 66.42 % of variability where two PCs components influenced mostly the variability (PC1 with 28.1 % and PC2 with 24.43 %). Accessions were grouped into three major clusters based on complete linkage, suggesting for a variance at the level of 27.50 % within a class and 72.50 % between classes. The results suggested that plant height, spike length, number of spikelet per spike were the most important characters in differentiating the genotypes.


INTRODUCTION
The Albanian Gene Bank has 3317 accessions of different crops in long-term storage (base collection), where wheat crop place an important role.Among 594 of wheat (Triticum durum Desf.and Triticum aestivum L.) accessions, approximately 270 accessions belong to bread wheat genotypes.The gene bank has the aim not only to preserve the germplasm but also to make available the plant resources into breeding programs, to improve cultivars or to develop new ones.
The evaluation of genetic variability based on morphological characters especially those of economic interest could also be used to select appropriate materials in breeding programs for crop improvement (Dos Santos et al., 2009).As previously reported (Al Khanjari et al., 2008), quantitative traits are often used to assess and describe the wheat characters due to their role in the estimation of genetic diversity and discrimination of closely related types.They were used to identify duplicates, to establish core collections in gene banks, to investigate relationships between landraces and their wild relatives, and for the most important tool, to prioritize material for use in breeding programs (Ariyo, 1993;Pecetti and Annicciarico, 1992).
Genetic diversity of wheat genotypes can be evaluated using morphological, which contribute toward grain yield as plant height, spike length, number of spikes per spike and grain (Maqbool et al., 2010).The correlation coefficient analysis is useful in the identification of characters that are positively correlated with yield (Maqbool et al., 2010;Bode et al., 2012).The evaluation of phenotypic variability by multivariate analysis gives the possibility to include a large number of accessions and to identify the most suitable resources for special traits (Goel et al., 2015).
Therefore the object of this study was the characterization of 92 accessions of bread wheat (Triticum aestivum L.), part of the base collection of the Albanian Gene Bank (Agricultural University of Tirana), in terms of diversity in morphological traits and association between each pair of these traits.

MATERIALS AND METHODS
The study was based on the characterization of the diversity of bread wheat (Triticum aestivum L.) germplasm.The plant material was characterized by a survey on land and laboratory, during the year 2016 in the experimental field of the Agricultural University of Tirana.The agronomic characters were measured after harvesting, using 20 plants from each accession.

Experimental site
The study was conducted at the Experimental Station of Institute of Plant Genetic Resources Valias, Tiranë.It lies at an altitude of 40 m above sea level and at Latitude 41°24'6.14"N and Longitude 19°44'9.93"E.

Methods (Experimental Design)
Experiment carried one replication during the autumn season 2016.During the crop year, the accessions were evaluated for different characters of quantitative type as: tiller capacity (TC), plant height/cm (PH), spike length/cm (SL), number of spikelet per spike (SpS), number of seeds per spikelet (GSp), number of seeds per spike (GS), seed size/mm (SeS) and of seeds per spike/g (WGS).Morphological characterization of the accessions was done according to international standards (IPGRI, 1985).

Statistical analyses
Statistical tests were carried out by the Statistical Package for Social Sciences (version 21) and JPM.

RESULTS
To obtain a successful breeding program, it is essential the information that researchers can get on the variability of germplasm within a crop spices.Morphological characterizing of the individual wheat accessions is useful in selection of the adaptable parents in the hybridization process.To assess the genetic diversity among 92 bread wheat germplasm, 8 quantitative traits were used and the estimated variation coefficient was high for agronomic traits as PH, SL, GS and WGS, similar with others authors (Ali et al., 2008;Sabaghina et al., 2014).Regarding PH trait it was observed a variation from 82.9 cm to 180.3 cm, for WGS among 92 accessions the minimum value measured was 0.28 g and the maximum 5.738 g, high variance resulted in GS trait (from ± 12.8 number of seeds per spike to ± 71, Table 2).AGB 0262 recorded the highest value for tiller capacity (± 3.8) whereas the 92 accessions presented an average of 2.56 for the same trait.Sabaghina et al. ( 2014) reported a higher tiller number (ranging from 1 to 6) measured at 56 bread wheat genotypes.Among the mean value of genotype for plant height trait, accession AGB 0268 recorded the highest mean value (±180.3cm) and genotype AGB 0258 resulted with the lowest plant height (±82.9 cm).
The results are higher from those reported by Sabaghina et al. (2014) plant height variation from 54.9 cm to 109.53 cm, whereas Mahmood et.al. (2006) obtained results ranging from 62 cm to 110 cm, while Aliu et.al. (2010) reported a range from 71 to 79 cm in different bread wheat genotypes.
The variation of plant height trait classified the 92 accessions in different classes (Table 3) where the major number of genotypes resulted from 91-100 cm.Similar results are reported by Peltonen et al. (2007) for the same trait.
Grain yield is influenced by spike properties and the spikelet number plays a very important role in the wheat grain yield (Sabaghina et al., 2014).Spike length in this study varied from 6.40 cm in AGB 0288 to 17.83 cm.
Results presented are higher from those reported from other authors (Peltonen et al., 2007;Sabaghina et al., 2014;Xhulaj et al., 2017).Comparing the mean values for SL and number of spikelet's per spike traits, the maximum values were observed in accession AGB 0251 (respectively 17.83 cm and 26.6 cm) followed by AGB 0268.Observations revealed that most of the wheat germplasm (50 accessions) were classified together for spike length trait measured between 9.1 to 11 cm (Table 4).6).
Regarding the number of grains per spikelet character, AGB 0322 had the highest mean value (4.0) and AGB 0266 recorded the lowest value (2.1).Among the 92 bread wheat genotypes 23 of the accessions recorded 2.0 to 2.9 number of GSp while 69 of the wheat accessions resulted with 3.0 to 4.0 number of GSp.
According to Othmani et al. (2015) this trait is regarded as the main wheat yield component and an increased grain number has been produced by spikes per unit or more grains per spike due to a higher spikelet number.
Data revealed that wheat genotype AGB 0326 presented the highest mean values for two traits, number of seeds per spike (±71) and seed size trait (±9 mm).Regarding GS trait, 92 accessions were grouped in different classes, from 5.43 % presented with 12.0 to 20.0 numbers of grains per spike, till 44.56 % (3.0 to 40.0 GS).
According to Okamoto et al. (2013) the grain number and mass as two main components in wheat grain yield are determined at different times of the growing season.This author suggested that seed mass best-explained genotype by environmental interaction for wheat grain yield.The 92 accessions revealed a high variation regarding this trait, where 26.08 % of them recorded values from 0.2 g to 1.2 g (Table 5), and most of the wheat germplasm (58.69 %) recorded values from 1.3 g to 2.3 g of the same trait.Observation showed that only one accession, AGB 0285 recorded the highest value in mass of seeds per spike (±5.73 g).Seed mass parameter also is important in wheat increasing seed germination percent, seedling emergence, tiller capacity, spike density and yield (Bellatreche et al., 2017).
Seed size trait recorded a high variation from 3.00 mm to 9.00 mm, representing one of the main components of the wheat yield, and increasing grain size continues to be a major breeding target (Sabaghina et al., 2014).Among 92 wheat germplasm in 16.3 % of them seed size varied from 3.0 mm to 5.0 mm, whereas most of the accessions (60.86 %) presented values from 5.0 mm to 7.00 mm for the same trait (Table 5).AGB 0276 had the lowest values for two traits WGS (0.28 g) and seed size (3 mm) followed for this last trait by AGB 0278 and AGB 0221.

Correlation Coefficient Analysis
Correlation of morphological traits was calculated by studying the data of bread wheat germplasm (Table 7).
Correlations measure the interdependence between a pair of characters.Knowledge of correlation is required to obtain the expected response of other traits when selection is applied to the trait of interest in a breeding program (Maqbool et al., 2010).Plant height showed positive significant correlation with yield contributing traits as spike length (r = 0.560) and the number of spikelet's per spike (r = 0.305).
Same results are reported from previous studies (Maqbool et al., 2010;Xhulaj et al., 2017).While significant negative correlation is observed among TC and SpS (r = -0.358).The number of spikelet per spike had a positive correlation with spike length trait (r = 0.589).Whereas number of grains per spike had a significant positive relation with if grains per spike trait (r = 0.719), supported by other works (Khaliq et al., 2004;Xhulaj et al., 2017).

Principal component analysis
The average data was analyzed using principal component analysis.According to the data (Table 8), three principal components exhibited about 66.42 % of variability where two PC components influenced mostly the variability (PC1 with 28.1 % and PC2 with 24.43 %; Figure 1).The first PC was related with plant height, spike length, number of spikelets per spike and of seeds per spike traits (Table 9) giving about 28.1 % of the variability but poor in tiller capacity.In the second PC traits as number of seeds per spike, seed size and WGS contribute at the level of 24.43 % of variability.The third principal component exhibited positive effects for seed size and number of seeds per spikelet (13.88 %), and maximum variation was observed for tiller capacity, plant height and number of seeds per spikelet at fourth, PC but poor in SL, SpS, GS and WGS.Different authors (Escobar-Hernandez et al., 2005;Othmani et al., 2015) used principal component method for grouping of germplasm.In addition to cluster analysis, biplot has been applied to study relation among studied traits in a set of genotypes (Aghaee et al., 2010;Peterson et al., 2005;Yan and Fregeau -Reid, 2008).Biplot (genotype by trait) explained the percentage variance associated with each principal component obtained by drawing a graph between Eigen values and principal components number.
The biplot (Figure 1) suggest that the best or the incompatible wheat genotypes in most of the traits, since they had the longest distance from the origin for the two principal components were AGB 0251 (32  78) resulted with the lowest performance for the measured traits (Figure 1).
The vector view of the biplot suggest a strong positive correlation among traits as WGS and GS, GSp, SeS; between SpS and SL, PH; also among SeS and GSp, GS as indicated by the small obtuse angles between their vectors.The correlation between WGS and PH, SpS, SL; among SeS and SL; between GSp and SL, SpS; and finally GS and SL was near zero as indicated by the near perpendicular vectors.The vectors indicated by the near angle of approximately 180 degrees, suggest for the existence of a strong negative correlation between TC and SpS, SL and PH; also between SeS and PH.
Comparing the Eigen values for each factor using the minimum Eigen value criterion, there are 3 main PC with Eigen values > 1.00 (Table 9 and Figure 1) that influence the genetic variability among 92 wheat genotypes.PC1 showed 28.1 % of variability with Eigen value 2.24 in germplasm which then reduced gradually.After the fourth PC little variance was observed and it ended at 2.76 % with Eigen value 0.22.
From the graph (Figure 1) the maximum variation was present in the first PC.So the selection of genotypes with desirable characters from this PC will be useful for further breeding programs.

Cluster analysis
The 92 wheat accessions were grouped according to quantitative traits into three major clusters based on complete linkage, whereas each cluster was statically different from each other (Figure 2).Cluster 1 consisted of 41 accessions, cluster 2 of 32 wheat accessions and the third cluster with 19 accessions (Table 10).The results suggested that plant height, spike length, number of spikelet per spike were the most important characters in differentiating the genotypes.The use of principal component analysis (showing the largest contributor to the total variance) and correlation coefficient analysis in the wheat germplasm, simplify dependable classification of genotypes, the identification of the superior genotypes (considering the evaluation of mean values) and their relation with morphological traits with possibility expenditure in breeding programs.Identification of the most important quantitative agronomical traits in wheat can facilitate selection of any individual accession and of desirable traits (genes), increasing the information of the wheat germplasm in gene bank.
The traits with more significant weighting on respective PC variance can be utilised successfully as quantitative markers for evaluation, characterization of the wheat germplasm stored in gene bank.Possible parental lines among these bread wheat genotypes that are in conservation in Albanian Gene Bank could be selected and utilised for sustainable field wheat breeding programs.

Table 1 :
List of the 92 wheat (Triticum aestivum L.) accessions object of the study

Table 2 :
Descriptive statistics of quantitative traits in 92 accessions of bread wheat (Triticum aestivum L.) TC -tiller capacity; PH -plant height/cm; SL -spike length/cm; SpS -number of spikelet per spike; GSp -number of seeds per spikelet; GS -number of seeds per spike; SeS -seed size/mm and WGS -of seeds per spike/g.

Table 3 :
Classification of 92 wheat (Triticum aestivum L.) accessions according to PH trait

Table 4 :
Classification of 92 wheat (Triticum aestivum L.) accessions according to SL trait

Table 6 :
Classification of 92 wheat (Triticum aestivum L.) accessions according to SpS trait

Table 8 :
Eigen values and percentage of total variance for PCA in 92 accessions of wheat

Table 9 :
Eigenvectors contribution in 92 accessions of wheat

Table 10 :
Clusters composition with 92 accessions (Code AGB) of bread wheat Dendrogram from cluster analysis of 92 bread wheat accessions based on quantitative traits Results of this study succeed in obtaining important scientific information on wheat germplasm database stored in the Albanian Gene Bank, and for further wheat breeding programs.The significant differences found in the present study show the existence of a high genetic variability among the 92 bread wheat genotypes and quantitative traits analysed, adequate for selection of desirable traits, and creation of new favourable gene combinations.Among the mean value of genotype for plant height trait, accession AGB 0268 had the highest mean value for SL and the number of spikelets per spike traits, the maximum values were observed in accession AGB 0251 followed by AGB 0268.Regarding number of grains per spikelet character, AGB 0322 had the highest mean value.Data revealed that wheat genotype AGB 0326 presented the highest mean values for GS and seed size trait.AGB 0276 had the lowest values for two traits WGS and seed size followed for this last trait by AGB 0278 and AGB 0221.Three principal components exhibited about 66.42 % of variability where two PCs components influenced mostly the variability (PC1 with 28.1 % and PC2 with 24.43 %).