Evaluation of Genetic Variability of Some Local Grass Pea ( Lathyrus sativus L.) Genotypes Using Different Statistical Analysis

Background: The aim of this study was to determine the variation among local grass pea varieties in Türkiye by using some statistical analysis. Methods: The investigation was carried out according to the randomized block design with three replications during the growing season of 2018 and 2019 with 9 different genotypes and 2 registered varieties of grass pea at the Bat ı Akdeniz Agricultural Research Institute. Days to flowering, green herbage yield, dry herbage yield, seed yield, thousand seed weight, plant main stem length, the number of pods per plant, the number of seeds per pod, straw yield, biological yield and harvest index were determined. Furthermore, the results of investigated features were subjected to correlation, regression, path and cluster analysis. Result: The results showed sufficient variability in seed yield based on characteristics among the tested genotypes. Regression analysis showed that days to flowering, harvest index, number of pods per plant, straw yield and biological yield should be given more importance during selection. This study resulted in three clusters with the maximum intergenetic distance between genotypes Ls17 and Ls23. According to the results of the research, it was determined that genotypes Ls15 and Ls32 were equivalent to commercial cultivars in terms of the investigated characteristics.


INTRODUCTION
Grass pea (Lathyrus sativus L.) is the most cultivated and economically important species of the genus Lathyrus in the W orld (Basaran et al., 2011).W hile there are 160 Lathyrus species all over the world, 61 of them are found in Türkiye (Genç and Sahin, 2001).Lambein and Kuo-Genth (1997) stated that there are archaeological remains showing that the Grass pea (Lathyrus sativus L.) was cultivated in the Balkans, Türkiye and Iraq in 8000 BC.Easily growing in different soil types and at high altitudes and growing with as little as 250 mm of annual precipitation, grass pea is typically the last surviving plant in drought periods (W hite et al., 2002).Moreover, it can be successfully grown in marginal areas in many parts of the world in severe climatic conditions without needing much production input (Arslan, 2019).Beyond its yield potential, grass pea has been stated to have remarkable yield stability in the face of drought (Zhelyazkova et al., 2016).This crop is mainly used as animal feed, however, it is occasionally used for human consumption in some parts of Türkiye because of the high protein content in its seeds (Zhao et al., 1999;Basaran et al., 2010;Atıs and Acıkalın, 2020).These positive characteristics make it a superior crop for food security, particularly in the face of anticipated climate challenges (Gonçalves et al., 2022).
According to Turkish Statistical Institute data, 1662 tons of prunes were produced from 14 820 decares of land in Türkiye and the grain yield was 112 kg da -1 (TUIK, 2021).The low yield potential of the genetic material used can be shown as the most important reason for the low yield of green herbage and seeds obtained from the unit area.In the breeding studies to be carried out to develop new varieties with high yield potential, the breeding criteria aimed for the breeding of suitable varieties should be well defined and the characteristics that need to be improved should be well known.Besides, the relationship between the characteristics and the mutual influences on each other need to be well-diagnosed.Finally, the relationships between these criteria can be decisive in making the choice (Oten et al., 2017b).The correlation coefficient is used to reveal the simple relationships between the examined features and the path analysis is used to find in detail the direct and indirect effects affecting the yield.It is known that the correlation coefficient does not always give definite results in determining the selection criteria (Cakmakci et al., 1998).

Evaluation of Genetic Variability of Some Local Grass Pea (Lathyrus sativus L.) Genotypes Using Different Statistical Analysis
The aim of this study was to determine the variation among local grass pea varieties in Türkiye and to determine the prominent characteristics as selection criteria in the breeding program by using some statistical analysis.

MATERIALS AND METHODS
The study was conducted in the trial area of the field crops department of Batı Akdeniz Agricultural Research Institute (BATEM) between 2018 and 2019 in Aksu (3694.077 N, 3089.008 E) in Antalya province in the Mediterranean region of Türkiye.In the first year of the experiment, sowing was planted on on October 17 and in the second year of the experiment was planted on October 9.The climatic data of the experimental area, average temperatures, rainfall and relative humidity belong to 2018-2019 and the long period are given in Table 1.
The long-period average precipitation of the experimental area was 1069.8 mm.During the experiment in 2018 and 2019, total annual precipitation was measured as 538.0 and 1097.0 mm respectively.While the total amount of precipitation in 2019 was similar to the long-term average, it was lower in 2018.The relative humidity was higher than the long-period relative humidity.Moreover, the average temperatures of 2018 and 2019 were similar with long period 9 (Table 1).
The soil of the research area was silty-clay loam with moderate organic matter and slightly alkaline pH.Salinity in the research area was 0.205 (dS m -1 ), pH 8.0, organic matter 2.17%, clay 31.0%and silica 50.0%(Table 2).In the research area, 4 kg N (Ammonium sulfate, 21%) and 8 kg P 2 O 5 (Triple super phosphate, 42%) per decare were applied with sowing every year.During the growing seasons, no chemical was applied, since there were no pests or diseases in the study.Moreover, weeds were controlled by hand hoeing.
Grass pea seed samples obtained from the previous project (Grass pea Breeding Studies) were used as material in this study.Nine candidate grass pea genotypes collected from the natural flora of Antalya province and selected as superior due to their high performance were used with two varieties as a control.Field experiments were conducted in the randomized block design with three replications.Each plot consisted of 6 rows and was 5 m in length.Sowing was made at the end of October in both years with 30 cm row spacing and a 10 kg da -1 seeding rate.In the study, days to flowering (day), green herbage yield (kg da -1 ), dry herbage yield (kg da -1 ), plant main stem length (cm), number of pods per plant, number of seeds per pod, seed yield (kg da -1 ), 1000 seed weight (g), straw yield (kg da -1 ), biological yield (kg da -1 ) and harvest index (%) were investigated.Morphological measurements were performed on ten plants randomly selected from each plot.The number of seeds per pod was evaluated on mature pods (5 pods per plant).All observations obtained in the study were taken according to the methods specified in TTSM (Anonim, 2001).Statistical analyses of the obtained data were performed by SAS ( 2002) statistical software.The treatment means were compared by Duncan's multiple range test (Duzgunes et al., 1987).The correlation, regression and cluster analysis were performed using the SPSS 16 statistical software and path analysis were performed by Path2 statistical software correlation and cluster analysis.

Mean values and Duncan groups in grass pea genotypes
The results of the Duncan groups are presented in Table 3.It is noteworthy that there was a large variation among genotypes in terms of days to flowering.Minimum and maximum values obtained from Ls36 (81.8 days) and Ls17 (119.2 days) are indicated in Table 3.Similarly, there was great variation among genotypes in terms of the amount of green herba yield.Another remarkable point is that the seed yield of the genotypes used in the research was high.It was observed maximum values on dry herbage yield values from Ls15 and Ls 32, thousends seed yield value from Gurbuz Based on the available data, it can be said that all genotypes were significantly different.
Since as a result of variance analysis the interaction of years and genotypes*years was insignificant; Correlation, Regression, Path and Cluster analysis were carried out with the means of two years values.

Correlation coefficients
The correlation coefficients between pairs of investigated traits are presented in Table 4. High correlations were    observed between green herbage yield and plant main stem length, straw yield, biological yield and dry herbage yield; biological yield with harvest index, seed yield, plant main stem height, number of pods per plant, number of seeds per plant pod and straw yield.On the other hand, the correlation was not determined on thousand seed weight with other investigated traits (Table 4.) Sayar et al., (2013) observed a positive relationship between hay yield and green forage yield and a significant and negative relationship between hay yield and 50% flowering days.In another study, a positive and significant relationship was determined between seed yield and green herbage yield, dry herbage yield, straw yield, biological yield and harvest index (Oten et al., 2017a).

Regression analysis
In the breeding studies, it is of great importance to know the effective factors on yield and to reveal the relations between these factors an d to d irect the breeding programs acco rdingly (Oten et al., 2017a).How ever, not all observations considered in a study have a direct effect on the dependent variable as an independent variable (Khomari et al., 2017).With the regression analysis, the effect type and degree of each independent variable on the dependent variable is determined and also the possibility of estimating the dependent variable arises (Andales et al., 2007).
In the regression analysis, days to flowering, harvest index, number of pods per plant, straw yield and biological yield were determined as remaining characteristics (Table 5).

Path analysis
The remaining characteristics in the stepwise regression model were determined and path analysis was performed with the data belonging to the characteristics that were determined to have a direct effect on seed yield (Table 6).The study of path analysis indicated that the highest positive direct effect towards seed yield was exhibited due to biological yield (0.848) and harvest index (0.409).This finding indicates that the selection for these characters is likely to lead directly to a full improvement in seed yield.The features that have an indirect positive effect on seed yield through biological yield are harvest index and the days to flowering and the features that have an indirect negative effect are the number of pods per plant and straw yield.According to the path analysis results of Hakyemez (2000), it was reported that the highest direct effect on grain yield was provided by dry herb yield.In the study by Oten et al. (2017a), it was stated that biological yield provided the highest direct effect on grain yield and although there was a positive and significant correlation between grain yield and thousand-seed weight, the indirect effect was negative.

Cluster analysis
In this study, regression analysis existing between investigated characters suggested the need for cluster analysis in association with seed yield.Besides, cluster analysis is often used to assess genetic diversity and to classify species (Van Beuningen and Busch, 1997).The clusters that were done according to Ward Method were illustrated by a dendrogram (Fig 1).Cluster analysis was performed to provide that genotypes were exactly screened based on the investigated traits.
The genotypes were divided into three gro ups according to the clustering results with the remaining traits as a result of the regression analysis.Genotypes Ls17, Ls33 and Ls41 were in the first group.W hile the second group included only the Ls25 genotype, the third group included Ls15, Ls23, Ls31, Ls32 and Ls36 genotypes and Gurbuz and Corea varieties.Evaluation of Genetic Variability of Some Local Grass Pea (Lathyrus sativus L.) Genotypes Using Different Statistical Analysis Furthermore, Ls17, which is one of the genotypes in the 1 st group, was found high in terms of days to flowering, while it was determined between the second and third groups in terms of other characteristics.It is seen that genotype 25 in the second group had a lower value than the other genotypes in terms of all the examined traits, except for the days to flowering.It was determined that the genotypes in the 3 rd group had higher values than the other groups in terms of the number of pods per plant, straw yield and biological yield features.Khosravi et al., (2022) stated that cluster analysis divided the genotypes into two separate groups, while Ahamed et al., (2012) divided the genotypes into ten separate groups in their study.

CONCLUSION
The results of this study, which was carried out to determine the variation among some local grass pea cultivars in Türkiye by using different statistical analyzes, showed that genotypes used in the study were superior in terms of all the traits and all the investigated characteristics were statistically significant.Although a low seed yield value was obtained from control cultivars, it was determined that genotypes had high values.It can be said that the Ls36 genotype was the earliest variety with 81.8 flowering days and this variety can be recommended for second-crop.Among genotypes, high correlations were observed between green herbage yield and plant main stem length, straw yield and biological yield.Furthermore, according to the results of the regression analysis, days to flowering, harvest index, number of pods per plant, straw yield and biological yield were determined as remaining characteristics.According to the results of the cluster analysis, it was determined that Ls15 and Ls32 genotypes in the third group were equivalent to commercial cultivars in terms of the investigated characteristics.

Table 1 :
Climate data (2018 and 2019/Long period).Evaluation of Genetic Variability of Some Local Grass Pea (Lathyrus sativus L.) Genotypes Using Different Statistical Analysis variety, plant main stem length values from Ls15 and Ls 32, number of pods per plant value from Ls36, number of seeds per plant pod values from Ls41, straw yield value from Ls15, biological yield value from Corea variety and Ls15 and harvest index values from Corea and Gurbuz varieties.

Table 3 :
Combined (2018 and2019)mean values and Duncan groups of the investigated characters of grass pea genotypes.

Table 2 :
Some physical and chemical characteristics of the soils *BATEM, 2019.

Table 4 :
Correlation coefficients between selected observations values of grass pea.
FD: Days to flowering, HI: Harvest index, NPP: Number of pods per plant, SW Y: Straw yield, BIY: Biological yield.

Table 6 :
Path analysis results.
FD: Days to flowering, HI: Harvest index, NPP: Number of pods per plant, SW Y: Straw yield, BIY: Biological yield/ Residual effects 0.045.