Variation for seed protein and ODAP content in grass pea ( Lathyrus sativus L.) germplasm collections

Development of grass pea ( Lathyrus sativus L.) varieties with low seed anti-nutritional factor, β -N-Oxalyl-L- α , β - diaminopropionic acid (ODAP) content but with high seed protein content would be beneficial for human consumption. In this study, a total of 702 germplasm accessions, from the International Center for Agricultural Research in the Dry Areas (ICARDA), Syria, were grouped into seven sets of trials based on their origin, and evaluated for seed ODAP content and seed protein content at two experimental sites, Tel Hadya and Breda over eight years. Significant genotypic differences were found for both the traits in all the germplasm sets except for seed ODAP content in BANG1 and ETH1. The effects associated with genotype × year within locations interactions were larger than genotypic effects for seed protein content in all the germplasm sets except ICARDA and for seed ODAP content in all the sets except ICARDA and PAK. The highest range was found for seed protein content (28.82-30.72%) in ICARDA germplasm set and for seed ODAP content in ETH2 (0.32-0.47%) and PAK (0.38-0.53%) germplasm sets. On the basis of best linear unbiased predictor values, new promising sources with low seed ODAP content such as ILG468, ILG1934, ILG1950 and ILG1951 and for high protein content such as ILG311, ILG670, ILG688, ILG691 and ILG708 and were identified for future grass pea breeding.


Introduction
Grass pea (Lathyrus sativus L.) is a cool season legume crop.It is a diploid (2n=14) (Talukdar 2009).It has high amount of protein content in seeds (27%) (Hanbury et al. 2000).It is a hardy legume crop; it can tolerate drought, moderate level of soil salinity and water-logging.It is currently cultivated in many May, 2019] Protein and ODAP content in grass pea 439 characterization of grass pea revealed the genetic diversity for these two quality traits (Bisignano et al. 2002;Granati et al. 2003;Tamburino et al. 2012).With this background, the present work was carried out to: 1) evaluate grass pea germplasm from Bangladesh, Pakistan, Nepal, Syria and Ethiopia at different locations over several years for seed protein and ODAP content, 2) assess the genetic variability and heritability for seed protein and ODAP contents, and 3) identify promising sources of genetic variation for future breeding.

Materials and methods
A total of 702 grass pea germplasm accessions from the seed bank of International Center for Agricultural Research in the Dry Areas (ICARDA), Aleppo, Syria were grouped into seven sets of trials based on their origin of collection, such as, "BANG1" (100 accessions) and "BANG2" (225 accessions), for the germplasm from Bangladesh; "ETH1" (100 accessions) and "ETH2" (64 accessions), for the germplasm from Ethiopia; "NEP" (49 accessions), for the germplasm from Nepal and "PAK" (64 accessions), for the germplasm from Pakistan.As ICARDA had its head quarters at Syria, the germplasm originated in Syria was named as "ICARDA" (100 accessions).Subsequently, all seven germplasm sets were evaluated for seed ODAP content and seed protein content in seven sets of trials at two experimental station of ICARDA namely, Tel Hadya and Breda over eight years (1998/99-2005/06).It is important to note that the accessions from a given set remained as same over years and locations.In each year, the crop growing season included the period from September to the following May.In each and individual trial, accessions were evaluated in simple lattice designs (Cochran and Cox 1957) , 1970) and seed ODAP content was determined calorimetrically using the ophtalaldehyde method of Rao (1978) which was further modified by Briggs et al. (1983).
Initially, the statistical analysis of traits data from individual trials was carried out while assuming a model with replication effects, block effects within replications and genotypes effects as "random".The test of equality of effects was carried out by fitting a mixed model using REML (Restricted Maximum Likelihood) approach.Further again, to examine how the response of genotypes in a given trial varied with year and location, a combined analysis of data over the location and year was carried out.Components of variance of genotype effects, genotype x location (G×L) interaction and genotype x year within locations (G×YwL) and pooled error were estimated by fitting a mixed effects model using REML.The model also assumed fixed effects for location, random effects of replications and blocks within replications within location -year combinations (i.e. the environments).In order to keep the estimates of variance components in a meaningful range, an option in the REML was set to produce only non-negative estimates of variance components.From the mixed models for the combined analysis the best linear unbiased predictor (BLUP) values were obtained for every individual trait.The broad-sense heritability (h 2 b ) was estimated using the following the formula: genotype x year interaction within location variance component, σ 2 e = pooled error variance, l = number of locations, y=number of years within a given location and r= number of replications.The heritability estimates were categorized as low (<0.30),moderate (0.31-0.60) and high (>0.60)(Robinson et al. 1949).All statistical analyses were carried out under GenStat software for windows, Release 18, VSNL International Ltd, Hemel, Hempstead, Hertfordshire, UK (VSN International. 2015).

Results and discussion
The REML for individual trials found significant variances due to genotypes (Data not presented).Further, the REML analysis of the combined data showed genotypic variability and interaction effect across locations (GxL) and year within locations (GxYwL) to variable degree and their significance is given in terms of P-values in Table 1.Significant genotypic (σ 2 g ) differences were found for seed protein content and seed ODAP content in entire germplasm sets except for seed ODAP content in BANG1 and ETH1.Incidence of significant genetic variation for [Vol.79, No. 2 seed protein content and seed ODAP content indicates great potential of these germplasm in future breeding programs (Table 1).
Significant G×L interaction effect were identified for seed protein content and seed ODAP content only in BANG1.But, significant G×YwL interaction effects were revealed for seed protein content in entire germplasm sets and for seed ODAP content in all germplasm sets except in BANG1 and PAK.Mostly, the effects associated with the G×YwL interaction were larger than the genotypic effects for seed protein content in all germplasm sets except ICARDA and for seed ODAP content in all germplasm sets except ICARDA and PAK (Table 1).Moreover, the same trait has also obtained a low, moderate and high level of heritability and confirmed again the major influence of climatic conditions of the growing seasons in Breda and Tel Hadya across the years (1998/99-2005/06) (Table 1).When compared the two experimental sites, Tel Hadya (360 mm) recorded higher average total growing season (September to June) rainfall than Breda (291mm) across the eight years.The range of total growing season rainfall (from September to the following May) was 261-483 mm at Tel Hadya and 198-386 mm at Breda.These variation in growing season rainfall from one year to the next even within the same location would have affected the crop response and hence the ODAP concentration in grass pea seeds.As has been stated in Piergiovanni and Damascelli (2011) the climatic factors such as rainfall and average temperature during the crop growing season largely affected the amount of anti-  The mean, range and standard deviation of various germplasm sets are presented in Table 2.The mean seed protein content in Bangladesh (28.95%) and Ethiopian (29.18%) germplasm are observed higher than the previous results by Kaul et al. (1982) and Urga et al. (1995).On the other hand, the germplasm from Syria (ICARDA) recorded lower the mean seed protein content (29.43%) than the earlier findings by Aletor et al. (1994).Among all germplasm sets, the highest range was found for seed protein content (28.82-30.72%) in ICARDA material but with low level of variation (CV = 2.0% in the predicted mean values).The problem of low level of variation (CV=2.58%)for seed protein content in Turkish grass pea landraces was stated by Basaran et al. (2013) in recent times.
In contrary to seed protein content, a relatively higher level of variation for seed ODAP content was revealed across all germplasm sets (Table 2).Among (ETH2) and 0.11 (ETH1) were non-significant at 5% level.The presence of non-constant correlation indicates the effect of natural selection that changes the genetic constitution of the accessions and their trait expression within each germplasm group which was obtained from different countries.
The five most desirable accessions selected based on BLUP values for seed protein content and seed ODAP content are presented in Table 3 and 4 respectively.All five selected accessions for seed protein content in BANG1, ETH2, ICARDA, NEP and PAK recorded significantly more protein than the local check.Particularly, accessions namely ILG311, ILG670, ILG688, ILG691 and ILG708 in ICARDA had the protein content of more than 30% (Table 3).Similarly, the accession ILG468 in ETH1, all five selected accessions in ETH2 and ICARDA, accessions ILG1934, ILG1950 and ILG1951 in NEP and all five selected accessions in PAK recorded significantly lower ODAP content than the local check (Table 4).The promising accessions identified in the all sets, the highest range was found in ETH2 (0.32-0.47%) and in PAK (0.38-0.53%).As shown by Kumar et al. (2011), our findings also agreed that the germplasm material from Ethiopia possess the highest variability (CV = 7.3%) for seed ODAP content than the germplasm from Bangladesh, Nepal and Pakistan (Table 2).The correlations between protein and ODAP content, based on their BLUPs for the accessions, varied with the trials: 0.57 (P<0.001,BANG1), 0.53 (P<0.001,ICARDA), 0.34 (P<0.001,BANG2), -0.30 (P<0.05,PAK), while the estimates, 0.26 (NEP), -0.19 present study could be used as parents in grass pea breeding programmes.
As these two traits are largely influenced by G×YwL interaction, it is also important to note that the success of breeding safe cultivars of grass pea is likely to be depending on the development of genotypes that express low levels of ODAP and high protein content in the given environments.Furthermore, the less genetic variability for seed protein content and the lack of genotypes with low components due to genotype (G), genotype ×location (G×L) interaction, genotype × year within location (G×YwL) interaction, and plot-errors.*, ** and **** significant at 5%, 1% and 0.1% respectively.S.E.= Standard Error.a = Variance components estimates kept at the boundary when restricted to positive range, b = Data not availableMay, 2019]    Protein and ODAP content in grass pea 441 nutritional factors accumulation in grass pea seeds.

Table 1 .
Estimates of components of variance and their standard errors, broad sense heritability of seed protein and ODAP content

Table 2 .
Summary of trial means, standard deviation and the range of the different quantitative traits based on the best linear unbiased predictor (BLUP) values of the accessions from the seven sets of trials Coefficient of Variation between the BLUP estimates (Standard deviation/PMean)