29-6-1006907.pdf

Online ISSN: 2345-475X University of Tehran Press DESERT Home page: https://jdesert.ut.ac.ir/ Introduction of dry yield-related traits to screen low-irrigation tolerant ecotypes in alfalfa (Medicago sativa L.) Sayyed Saeed Moosavi1* , Leyla Ahmadi1 , Mohammad Souri2 , Mohammad Reza Abdollahi1  1* 2 Department of Plant Production and Genetics, Faculty of Agriculture, Bu-Ali Sina University, Hamedan, Iran. E-mail: [email protected] Seed and Plant Research Department, Hamedan Agricultural and Natural Resources Research and Education Center, Hamedan, Iran. Article Info. Article type: Research Article Article history: Received: 07 March 2024 Received in revised from: 16 June 2024 Accepted: 22 June 2024 Published online: 27 June 2024 Keywords: Alfalfa, Factor analysis, Cluster analysis, Discriminant function analysis, Stepwise regression, Causality analysis. ABSTRACT Alfalfa is one of the most important forage crops in the world and Iran. Due to the adverse effect of drought on alfalfa yield, screening drought- tolerant genotypes is essential in breeding efforts. In the present study, 11 alfalfa ecotypes were evaluated during two years under the low-irrigation stress condition. The statistical analyzes were done on the average of two-years data. The first and second factors, respectively as "forage- quantity factor" and "foragequality factor", explained 70.40% of the data total variance. Factor analysis showed that, the traits of fresh-forage yield, plant height, stem fresh weight and regrowth rate, had the most positive effect on dry-forage yield, respectively. Qharaghezlou ecotype with the highest dry-forage yield, and Sedghiyan ecotype, with the highest forage quality, were the most drought-tolerant and high-quality ecotypes, respectively. Ecotypes of Mohajeran and Famenein showed the lowest dry-forage yield. Ecotypes were grouped in three separate clusters. The first and third clusters were identified as "dry-forage quality" and "dry-forage quantity" cluster, respectively. These two clusters had the highest genetic distance. Correctness of cluster grouping was confirmed by the discriminant function analysis. Fresh-forage yield, dry to fresh-forage yield ratio, plant height and regrowth rate were entered into the regression model respectively, as the most important traits affecting on dry-forage yield. The traits of fresh-forage yield and dry to fresh-forage yield ratio showed the most positive direct effect on dry-forage yield. Also, plant height, through increasing fresh-forage yield, and regrowth rate, through decreasing the dry to fresh forage yield ratio, had the largest positive and negative indirect effects on dry-forage yield, respectively. According to the results, the ecotypes showed a high diversity, which suggests the use of desirable traits and superior genotypes identified for use in future alfalfa breeding programs. Cite this article: Moosavi, S.S., Ahmadi, L., Souri, M., Abdollahi, M.R. (2024). Introduction of dry yield-related traits to screen low-irrigation tolerant ecotypes in alfalfa (Medicago sativa L.). DESERT, 29 (1), DOI: 10.22059/jdesert.2024.98234 © The Author(s). DOI: 10.22059/jdesert.2024.98234 1. Introduction Publisher: University of Tehran Press 90 DESERT, 29-1, 2024 Unfortunately, the significant warm weather in recent years may increase the severity of drought, which is due to the decrease in rainfall and the increase in evaporation and transpiration in some regions of the world (Solomon et al., 2007). Drought stress is the most important abiotic stress that reduces crop yield in arid and semi-arid regions, such as Iran (Nazari et al., 2018). Based on the expansion of the range of this stress, conducting basic research to identify low-irrigationtolerant ecotypes is a basic, economic and necessary solution (Ranjbar et al., 2015; Neupane and Guo, 2019). Although various researches have been carried out in the field of breeding to improve tolerance to drought stress, but due to the significant interaction between genotype and environment, the process of these researches has been relatively complicated and slow. Medicago sativa L. is a perennial plant, cross-fertilization with about 10% self-fertilization, from the legume family. Its primary origin is northwest Iran, northern Turkey and the Caucasus region. This plant, as the queen of forage crops, can produce about two tons of protein per hectare of cultivation annually (Nekoyanfar et al., 2017). Considering the significant role of this plant in feeding livestock on the one hand and the spread of drought stress on the other hand, conducting basic research on alfalfa breeding to improve tolerance to drought stress is necessary and essential (Pour-Farhad et al., 2008). The presence of high genetic diversity and selection is a requirement for success in breeding a species (Moosavi et al., 2013). In this regard, multivariate statistical methods are useful methods to obtain efficient information in evaluating the diversity between different genotypes to identify superior genotypes and desirable traits. Among the various multivariate methods, principal component analysis, factor analysis, regression analysis, cluster analysis and causality analysis are widely used to separate the genotypes (Moosavi et al., 2022). Multivariate statistical methods, especially the regression method and path analysis can be used to investigate the real relationships between the traits (Le-Singer et al., 2005). Previously, these methods have been effectively used to understand the relationship between yield and agro-morphological traits affecting yield in crops. Le-Singer et al. (2005) used cluster analysis to group alfalfa genotypes using Ward's method. Bahar et al. (2005) classified alfalfa genotypes into 6 groups using cluster analysis to investigate Iranian alfalfa genetic diversity. Considering that some diverse agro-morphological traits have high heritability and can be easily measured, therefore, selecting based on them will be appropriate for improving yield (Julier et al., 2000; Sharma and Hore, 1993). Due to the complexity of yield, it is possible that sometimes indirect selection to improve yield, through yield-related traits with high heritability, will lead to better selection response (Jomova et al., 2005, Moosavi et al., 2020). Researchers believe that because the average of several samples has a better estimate of the parameter, therefore it is better to carry out repeated sampling and use the average of the samples to analyze the data (Farshadfar et al., 2007). By identifying genotypic groups with the greatest genetic distance through cluster analysis, it is possible to identify different parents for future crosses. So that different parents produce more heterosis and increase the probability of obtaining superior differentiated progeny through the phenomenon of transgressive segregation. On the other hand, determining the characteristics and grouping of germplasm allows breeders to avoid re-sampling of populations (Sharma and Hore, 1993). Despite the fact that the main efficiency of cluster analysis is to identify real groups and reduce the volume of data, sometimes it is possible that this statistical method creates unexpected groups. In this case, they should be analyzed through the decomposition of the discriminant function (Moghadam et al., 1995). Analysis of the discriminant function is used when certain groups already exist and the goal is to test the difference between these groups (Rabiei et al., 2008). In general, multivariate statistical methods are a common method for identifying the most important traits affecting yield and determining the relative contribution of each of them on Determination of erodible areas using MCDM models in …/ Chezgi et al. 91 total variance. The use of multivariate statistical methods in different plants has been reported by researchers. For example, some researches, such as Rabiei et al. (2008) in potato, AbdullahiMandolkani (2011) in alfalfa, Hazeq-Jafari et al. (2014) in alfalfa, Kakai and Mazahery- Laqab (2014) in alfalfa, Khodarahmpour and Motamedi (2016) in alfalfa, Tucak et al., (2009) in alfalfa and Moosavi et al. (2020 and 2022) in wheat, can be mentioned. The purpose of this research was to evaluate the genetic diversity and introduce the most desirable traits effective in increasing dry-forage yield to screen alfalfa ecotypes under lowirrigation conditions. 2. Materials and methods 2.1. Plant materials and experimental design In this research, 11 alfalfa ecotypes (Table 1) were evaluated in the form of randomized complete block design with 3 replications during 3 years. This plant germplasm is the best ecotypes of cold regions of Iran, which were compared and evaluated in this research. The first year was considered as the year of seedling establishment. Therefore, the traits were measured during the second and third year (two consecutive year). The land preparation consisted of turning plow, two perpendicular discs and a trowel, which was done before sowing the seeds. Then the required fertilizers were used based on the soil-test results. Then, rows with a distance of 50 cm were created using a furrow machine. Each experimental plot consisted of 6 rows with a length of 8 meters, and the distance between the rows was equal to 50 cm. Accordingly, the area of each plot was 24 square meters. The distance between plots was 50 cm and the distance between replicates was 250 cm. Planting was done manually in the beginning of September and immediately after planting, irrigation was done. Irrigation was carried out in a leaky manner. Low-irrigation stress was done by increasing the length of the irrigation period from 7 days to 12 days. This stress was applied after the establishment of seedlings and the development of their roots. During the experiment, the fight against pests and diseases was carried out. Then, different traits were measured in four cuts in the second and third years of planting. Table 1. Information of 11 alfalfa ecotypes used in the experiment Ecotype code 1 2 3 Ecotype name Ghahavand Ordoubad Chaleshtar Ecotype code 4 5 6 Ecotype name Gharaghezlou Gharghelough Sedghiyan Ecotype code 7 8 9 Ecotype name Malek-kandi Hamedan Famenein Ecotype code 10 11 --- Ecotype name Galebani Mohajeran --- 2.2. Trait measurement and data collection Recording of the traits (Table 2) was done during the second and third year of planting in four cuts in each year. The time interval between greening in spring and when 50% of the plants in each plot have at least one flower was considered as the trait of days to flowering. Before forage harvest, 10 plants were randomly selected from each experimental plot and plant height was recorded in centimeters. After removing 50 cm from the beginning and end of each row and after removing the marginal rows at the time of harvesting, the bushes of the remaining rows were harvested and immediately the fresh forage yield per area unit was recorded. Then, the above samples were kept for 48 hours in an oven with a temperature of 75 degrees Celsius and the dry forage yield per area unit was recorded. 250 grams of fresh forage samples were randomly selected in each plot at the time of harvesting and after separating the stems and leaves, the stem and leaf fresh weight was recorded. At the time of harvesting, from each experimental plot with 92 DESERT, 29-1, 2024 a length and width of 50 cm, it was harvested separately and the stems number per unit of 0.25 square meters was counted and then converted to the number of stems per square meter. Stems less than 20 cm were not counted. To record the regrowth rate, 17 days after harvesting each cut, that is, half the time of each cut, the height of 10 plants was randomly measured. Table 2. Information of the measured traits in the present experiment Trait code 1 2 3 4 5 Trait name (unit) Days to flowering (day) Plant height (cm) Dry forage yield (area unit) Fresh forage yield (area unit) Dry to fresh forage yield ratio Trait code 6 7 8 9 10 Trait name (unit) Stem fresh weight (g) Leaf fresh weight (g) Leaf to stem fresh weight ratio Stem number per square meter Regrowth rate 2.3. Data analysis Statistical analysis was done based on two-year average data. These analyzes included factor analysis, cluster analysis, discriminant function analysis, stepwise regression analysis, and causality analysis, which were performed using SPSS and Minitab software. Factor coefficients were calculated based on the principal components. Cluster analysis was done using Minitab software based on Pearson's squared distance matrix and Ward's method for grouping ecotypes. The discriminant function was analyzed to determine the difference between the groups using SPSS software. Stepwise regression analysis was performed using the forward method to identify traits with the greatest effect on dry forage yield and determine the relative contribution of each trait using Minitab software. Causality analysis in order to determine the cause-andeffect relationship of traits affecting dry forage yield, and to evaluate their direct and indirect effects using Path software. 3. Results 3.1. Factor analysis results The first and second factors explained 45.3% and 25.1%, respectively, and together 70.40% of the total data variance (Table 3). The first factor showed a positive and strong relationship with the traits of stem fresh weight, fresh forage yield, plant height, regrowth rate and dry- forage yield (Table 3). Accordingly, the first factor was named "dry-forage quantity factor". Meanwhile, the second factor showed a strong and positive relationship with the traits of leaf fresh weight, stem number per square meter, and leaf-to-stem fresh weight ratio. Based on this, the second factor was named "dry-forage quality factor". Fresh forage yield Dry forage yield Plant height Dry to fresh forage yield ratio Stem fresh weight Leaf fresh weight Leaf -to- stem fresh weight ratio Stem number per square meter Regrowth rate 45.30 25.10 70.4 Days to flowering 1 2 Sum Variance (⸓) Factor Table 3. The results of the coefficients of the first and second factors for different traits under low-irrigation conditions -0.38 -0.72 0.78 -0.42 0.49 -0.77 0.71 -0.21 -0.43 -0.78 0.84 -0.06 -0.84 0.60 -0.88 0.45 0.06 0.49 0.64 0.26 Determination of erodible areas using MCDM models in …/ Chezgi et al. 93 Fig. 1. The biplot of the first and second factors for different ecotypes and traits under low-irrigation conditions. 3.2. Cluster analysis results According to the results (Figure 2), the studied ecotypes were placed in three separate clusters. According to the grouping of ecotypes in the first and third clusters, these two clusters were identified as "forage quality cluster" and "forage quantity cluster", respectively. Clusters No. 1 and 3 showed the greatest genetic distance from each other (Table 4). Figure 2. Cluster analysis for different ecotypes and traits under low-irrigation conditions. Table 4. Distances between cluster centroids Cluster 1 Cluster 2 Cluster 3 Cluster 1 0.00 3.75 3.93 Cluster 2 3.75 0.00 3.85 Cluster 3 3.93 3.85 0.00 94 DESERT, 29-1, 2024 3.2. Discriminant function results The purpose of analyzing the discriminant function is to investigate the differentiation between groups in terms of several variables. Also, one of its goals is to verify cluster results (Farshadfar, 1998). The results of the analysis of the discriminant function confirmed the correctness of the grouping obtained from the cluster analysis one hundred percent (Table 5). Table 5. The results of the discriminant function to evaluate the correctness of the grouping obtained from the cluster analysis of 11 alfalfa ecotypes Number of members in each group Approval percentage of members of each group Number of members in each group Approval percentage of members of each group Total Approval percentage of members of each group 3 Number of members in each group 2 Approval percentage of members of each group 1 Number of members in each group The groups resulting from cluster analysis Predicted groups based on the discriminant function 1 4 100 - - - - 4 100 2 - - 2 100 - - 2 100 3 - - - - 5 100 5 100 3.3. Stepwise regression results Stepwise regression was used to determine the traits with the greatest effect on the two-year average yield of dry-forage and then to determine the contribution of each trait in the variance of dry forage yield (Table 6). Fresh-forage yield, dry to fresh forage-yield ratio, plant height and regrowth rate, as the most important traits affecting dry-forage yield, were entered into the regression model in order and explained 99.52% of changes in dry forage yield. Table 6. Stepwise regression analysis for dry yield, as dependent variable, and other traits, as independent variables, in 11 alfalfa ecotypes Regression coefficient for traits 1- Fresh 2- Dry to fresh 3- Plant forage yield forage yield ratio height 1 309.4 0.215 2 -906.1 0.221 45.10 3 -1827.7 0.163 56.50 15.60 4 -2014.4 0.155 65.40 12.40 **; indicates significance at the one percent level. Regression step Constant 4- Regrowth rate 11.20 Cumulative detection coefficient (R2) 73.50** 93.51** 98.40** 99.52** 3.4. Causality analysis (path analysis) results According to the obtained results, two traits of fresh forage yield and dry to fresh forage yield had the most positive direct effect on dry forage yield. Also, plant height traits, through increasing fresh forage yield, and regrowth rate trait, through decreasing the dry to fresh forage yield ratio, had the largest positive and negative indirect effect on dry forage yield, respectively (Table 7). Determination of erodible areas using MCDM models in …/ Chezgi et al. 95 Table 7. The results of causality analysis for dry yield with traits entered in the step-by-step regression model Trait Fresh yield Dry to fresh yield ratio Plant height Regrowth rate Residual effect= 0.132 Direct effect of traits on dry yield 0.654 0.636 0.247 0.173 Indirect effect of a trait on dry yield by other traits Fresh yield Dry to fresh yield ratio Plant height Regrowth rate Correlation with dry yield ---0.035 0.452 0.312 -0.034 ---0.232 -0.404 0.171 -0.091 --0.171 0.082 -0.111 0.119 --- 0.873 0.599 0.586 0.252 4. Discussion 4.1. Factor analysis discussion The obtained results (Figure 1, Table 3) indicate that the selection of ecotypes based on high values of the first factor will lead to the selection of ecotypes with high dry forage yield. Meanwhile, the selection based on the high value of the second factor will lead to the selection of high-quality forage genotypes. Generally, there is a negative relationship between quantity and quality of forage. That is, the genotypes that have a high performance usually have a lower forage quality. Considering the positive and negative relationship between forage yield and the first factor and the second factor, respectively, a high amount of the first factor and a low amount of the second factor are desirable. Based on this, the fourth region of the bi-plot was chosen as the most desirable region of the bi-plot (Figure 1). The ecotypes located in the fourth region, including the Qharaghezlou ecotype, had the highest fresh and dry forage-yields under low-irrigation conditions (Figure 1). This ecotype was selected as the most favorable genotype tolerant to drought stress. Meanwhile, if the goal of choosing ecotypes is based on forage quality, the most desirable area of the bi-plot is the second area of the bi-plot. Sedghiyan ecotype, as an ecotype with the highest forage quality, is located in the second area of the biplot diagram (Figure 1). According to the results of the factor analysis, the traits of fresh forage yield, plant height, stem fresh weight and regrowth rate, respectively, had the most positive correlation with the increase of dry forage yield (Figure 1). Qharaghezlou ecotype with the highest dry forage yield, and Sedghiyan ecotype, with the highest quality of forage, were screened and identified as the most drought-tolerant and high-quality ecotypes, respectively. Meanwhile, the two ecotypes of Mohajeran and Famenein showed the lowest dry forage yield. Kakai and Moosavi (2016), in a study of 19 chickpea genotypes, identified three factors of “grain yield", "harvest index", and "plant vigor". These three factors together explained 68.8% of the total variance. In research by Kakai and Mazahery-Laghab (2014), it was reported that the first three factors together explained 65.70% of the total data variation. Hazeq-Jafari et al. (2014), the first five factors explained 77.02 percent of the total variance. Farshadfar et al. (2007) also stated that 70/69 of the data variance is explained by the first three factors. They reported that plant height traits, wet forage yield and dry forage yield showed the highest correlation with the first factor. Ba-Safa and Taheryan (2006) in the study of 21 ecotypes of alfalfa, separated the number of 22 variables into 2 factors, one of which was called the quantity factor and the other was named the forage quality factor. Tucak et al., (2009) showed that the first four principal components contributed to 89.02% of the total variability among the populations and cultivars. The yields of green mass and dry 96 DESERT, 29-1, 2024 matter, vigour, growth habit, plant regeneration and length of central leaflet were the most important traits for the yield. Moosavi et al. (2020) reported that the first and second factors, as “yield and yield-components” and “vegetative growth” factors, respectively, explained 76.4% of the data on the total variance. Sayed et al., (2022) showed that the principal analysis grouped the studied genotypes into groups and remained scattered in all four quadrants based on all studied traits. 4.2. Cluster analysis discussion Cluster analysis is one of the most important common statistical methods for evaluating the diversity, grouping and screening of genotypes based on different traits. This method plays an important role in plant breeding selection programs. According to the grouping of ecotypes in the first and third clusters, these two clusters were identified as "forage quality cluster" and "forage quantity cluster", respectively. According to the genetic distance of cluster 1 and 3 ecotypes, they can be used for crossbreeding program and the possibility of obtaining the desired diversity for interbreeding activities in future programs. Ba-Safa and Taheryan (2006), divided the number of 21 alfalfa ecotypes into 4 groups, and the second group had high plant height, high dry and fresh forage yield, and fast regrowth rate, and was identified as a desirable cluster. Pour-Farhad et al. (2008) separated different alfalfa ecotypes into 5 groups with the help of Ward's cluster analysis. Hazeq-Jafari et al. (2014), 49 alfalfa genotypes were grouped into three clusters. The placement of some foreign and domestic cultivars in a cluster was an indicator of the non-compliance of the existing genetic diversity with the geographical distribution. In research on 20 alfalfa genotypes, Khodarahmpour and Motamedi (2016) clustered the genotypes into three separate groups. In the first cluster, there were lines that were optimal in terms of forage yield. In the second cluster, there were lines that were low both in terms of quantity of forage and quality of forage. The third cluster included the cultivars, which were optimal in terms of forage quality. The results of cluster analysis (Abdollahi-Madoulakani, 2012), examined 110 genotypes of 11 different populations of alfalfa and placed them in two main groups, each group including two distinct subgroups. Kakai and Mazahery-Laghab (2014) classified 30 alfalfa populations into three groups. The greatest genetic distance in this study was observed between Karaj 2122 and Zabul 15 populations. It is expected that the crossing of the mentioned populations in the diverging generations will provide a favorable diversity for future breeding programs. Tucak et al., (2009) showed that the cluster analysis performed on the two first components grouped the majority of alfalfa cultivars and populations into three clusters of germplasm. In most cases, clustering did not depend on the country of origin. In research on wheat (Moosavi et al., 2020) showed that the bi-plot and cluster results revealed a remarkable genetic diversity among the genotypes; therefore, these results might be helpful to identify donor parents in wheat breeding crosses for yield increscent. In another research (Sayed et al., 2022) revealed that the genotypes F18, F49, K75, S35, P20, P5 and P13 showed the highest values for all studied traits compared with other clusters. 4.3. Discriminant function discussion According to the results, four ecotypes with low forage yield, two ecotypes with medium forage yield and five ecotypes with high forage yield were placed in the first, second and third clusters, respectively. The analysis results of the recognition function confirmed this grouping with 100% accuracy. Ahmad Khan, in the study of chickpea genotypes (2009), used detection function analysis to verify cluster analysis grouping, and the accuracy of grouping was confirmed to be about 80. In research (Moosavi et al., 2022), the results of the cluster analysis Determination of erodible areas using MCDM models in …/ Chezgi et al. 97 were confirmed 100% by the analysis of the discriminant function. According to the results of the discriminant function analysis by Khoda-Rahmpour and Motamedi (2016), the genotypes were 100% correctly assigned to the clusters resulting from the cluster analysis. 4.4. Stepwise regression discussion Stepwise regression analysis is a prerequisite method for causality analysis in order to identify the most important traits affecting dry-forage yield. In research (Khoda-Rahmpour and Motamedi, 2016), it was stated that the two traits of fresh-forage yield and days to flowering were entered into the regression equation as the most important traits affecting dry forage yield. Moosavi et al., (2022) reported that water use efficiency, fertile spikes number per plant, seed number per the main spike, biological yield per plant, and water use (with a negative regression coefficient), as the most important traits, were entered into the regression model, respectively. Moosavi et al., (2020), reported that in a stepwise regression, the traits of fertile spikes number/ m2, spike weight/m2, grain number per spike and plant harvest index, as the most effective variables, explained 94.06% of grain yield variance in wheat. 4.5. Causality analysis discussion The purpose of causality analysis is to provide a proper analysis of the correlation between variables by creating cause-effect models. To identify ecotypes with high yield, it is necessary to identify traits that have a significant relationship with dry forage yield. Then, by selecting these traits, he proceeded to collect desirable genes in breed cultivars. Although correlation coefficients are widely used in identifying trait relationships, sometimes these relationships may be misleading. In fact, the high correlation between two traits may be the result of indirect effects of other traits, on the basis of which, the use of simple correlation analysis cannot explain the relationships between traits (Ali et al., 2003). Therefore, it is necessary to separate the direct and indirect effects of effective traits on dry-forage yield. Causality analysis can separate the direct effects of each yield-component on the final yield from the indirect effects (Emam and Borjan, 2000). For this reason, plant breeders use the causality analysis method as a tool to determine the importance of traits affecting yield. Considering that the correlation coefficient between dry-forage yield and fresh-forage yield and dry to fresh-forage yield ratio was approximately equal to the causation coefficient (direct effect), in this case the correlation coefficient indicates the real relationship between the two variables. Therefore, direct selection through these two traits can be useful in increasing the dry-forage yield of alfalfa. Therefore, based on the results obtained in this research, selecting cultivars with higher forage yield and dry to fresh forage yield ratio will increase the dry forage yield of the plant. In research on wheat (Moosavi et al., 2022) the most direct effect on increasing grain yield was water use efficiency. Seed number per plant and fertile spikes number per plant, due to increased water use efficiency, showed the most indirect effect on grain yield. Fertile spikes number/m2 and spike weight/m2 (Moosavi et al., 2020) revealed the maximum direct and positive effect on grain yield enhancement. The first and second factors, as “yield and yield-components” and “vegetative growth” factors, respectively, explained 76.4% of the data on the total variance. 5. Conclusion According to the results, the ecotype of Qharaghezlou was identified as the most favorable ecotype tolerant under the conditions of low irrigation stress. Also, the increase of traits such as fresh-forage yield, plant height, dry to fresh-forage yield ratio and regrowth rate are desirable traits for selection for moisture stress conditions. The most desirable ecotype in terms of quality was the Sedghiyan ecotype, which was located in the second area of the bi-plot and in the 98 DESERT, 29-1, 2024 forage-quality cluster. Two traits of fresh forage yield and dry to fresh-forage yield ratio had the most positive direct effect on dry-forage yield. Also, the plant height, through increasing the fresh-forage yield, and the trait of regrowth rate, through the reduction of the dry to freshforage yield ratio, had the largest positive and negative indirect effects on dry-forage yield, respectively. In general, the diversity of the mentioned ecotypes was confirmed based on various multivariate analyses, which can be used for future alfalfa breeding programs. Considering the significant genetic distance between the ecotypes of the first and third clusters, the crossing of these ecotypes can be used to create diversity in future alfalfa breeding programs. Acknowledgements We are grateful to Hamedan Agricultural and Natural Resources Research and Education Center, Hamedan, Iran. The study was supported by Bu-Ali Sina University grant number 2937, Hamedan, Iran. Conflict of Interest The authors declare no conflict of interest. References Abdollahi-Mandoulakani, B., 2011. Study of genetic diversity in some populations of cultivated alfalfa (Medicago sativa L.) using morphological traits. New Genetics. 7(4); 381-388. Ahmed-Khan, T., 2009. Application of univariate and multivariate techniques in evaluation of chickpea (Cicer aretinum L.) genotypes. M.Sc. Thesis. Department of Agriculture Statistics, College of Agriculture, Harward University of Agriculture Sciences. Ali, N., F. Javidfar, J. Elmira, M.Y, Mirzaee. 2003. Relationship among yield components and selection criteria for yield improvement in winter rapeseed (Brassica napus L.). Pakistan Journal of Botany. 35 (2); 167-174. Bahar, M., S. Ghobadi, V. Erfani-Moghadam, A. Yamchi, M. Talebi-Badaf, 2005. Evaluation of genetic diversity of indigenous populations of Iran alfalfa using ESTS microsatellite markers. Journal of Agricultural Sciences and Techniques, 2; 141-151. Emam, Y., A.R. Borjan, 2000. Yield and yield components of two winter wheat cultivars in response to rate and time of foliar application. Journal of Agriculture Science, 2: 263-270. Farshadfar, E. 1988. Application of quantitative genetics in plant breeding. Razi University Press. Page 186. Farshadfar, M., Sh. Fareghi., E. Farshadfar, A.A. Jafari, 2007. Evaluation of alfalfa genetic diversity using morphological and chemical indicators. Journal of Genetic Research and Breeding of Pasture and Forest Plants of Iran, 16(1); 1-13. Guertin, W.H., J.P. Bailey, 1982. Introduction to modern Factor Analysis. Edwards Brothers Inc. Michigan.405 P. Hazegh-Jafari, H., S. Aharizad, S.A. Mohammadi, F. Noormand Moayyed and P. Behrooz, 2014. Grouping of alfalfa genotypes based on different characteristics using multivariate statistical analysis. Journal of Crop Breeding, 6; 107-121. Jafari, A. A., M. Ziaei-Nesab, S.M. Hessamzadeh, H. Madahe- Arefi, 2004. Evaluation of genetic diversity of forage and seed yield in populations of red clover (Trifolium pretense L.) using statistical methods of multivariate analysis. Journal of genetic research and breeding of pasture and forest plants of Iran, 12(1); 91-109. Determination of erodible areas using MCDM models in …/ Chezgi et al. 99 Jeena, A.S., P.P. Arora, 2002. Multivariate technique in chickpea. Agric. Sci. Digest, 22(1): 57-58. Jomova, k., M. Benkova., M. Zakova., E. Gregova, J. Kraic, 2005. Clustering of chickpea (Cicer arietinum L.) accessions. Genetic Resources and Crop Evolution. 52; 1039-1084. Julier, B., C.H. Huyghe, C.H. Ecall, 2000. Within and among cultivar genetic variation in alfalfa: forage quality, morphology and yield. Crop Science. 4: 362-365. Kakaei, M., H. Mazahery-Laghab, 2014. Evaluation of alfalfa (Medicago sativa L.) germplasm using multivariate statistical analysis. Iranian Journal of Rangelands and Forests Plant Breeding and Genetic Research. 22(1); 125-132. Kakai, M. 2018. Studying the effect of genotype and drought stress on the physiological, morphological, phenological and biochemical characteristics of autumn rapeseed. Master's thesis in the field of agricultural engineering, Plant Breeding, Islamic Azad University. Kakai, M., S.S. Moosavi, 2016. Evaluation of genetic diversity and identification of traits affecting the yield of chickpea lines using multivariate statistical methods. Journal of Environmental Science, 15(2); 21-38. Keneni, G., E. Bekele., F. Assefa., M. Imtiaz., T. Debele., K. Dagne, and E. Getu, 2012. Phenotypic diversity for symbio-agronomic characters in Ethiopian chickpea (Cicer aretinum L.) germplasm accession. African Journal of Biotechnology, 33 (63); 12634-12651. Khodarahmpour, Z., M. Motamedi, 2016. Study of Genetic Diversity of Alfalfa (Medicago sativa L.) Genotypes Via Multivariate Analysis. Journal of Crop Breeding, 8(19); 163-169. Le-Singer, C., K. Gallardo, M.J. Prosper, C. Salon, L. Quillien, K. Thompson, 2005. Genetic diversity for seed protein composition in Medicago trunculata. Plant Genetic Resources, 3; 59-71. Moghadam, M., A. Mohammadi, M. Aghaie Sarbarze, 1995. Introduction to multivariate analysis methods. Parivar Press. 218 pp. Moosavi, S.S., F. Abdi, M.R. Abdollahi, S. Tahmasebi-Enferadi, M. Maleki, 2022. Identifying the most important traits affecting grain yield of wild wheat (Triticum boeoticum) under drought stress conditions. Desert, 27(2); 342-352. Moosavi, S.S., M. Nazaria, M. Chaichib, K. Jamshidi Goharrizi, 2020. The most effective yieldcomponents associated with increasing yield of wheat (Triticum aestivum L.) under terminal drought stress conditions. Desert, 25(2); 139-146. Moosavi, S.S., M.R. Abdollahi, F. Kianersi, 2013. The application of multivariate statistical methods in identifying traits affecting bread wheat yield under moisture stress conditions. Cereal Research, 2(3): 119-130. Nazari, M., Moosavi, S.S., M. Maleki, 2018. Morpho-physiological and proteomic responses of Aegilops tauschii to imposed drought stress. Plant Physiology and Biochemistry. 132; 445-452. Nekoyanfar, Z., Sh. Lack, Gh.R. Abadouz, 2017. Assessment effect of cutting time and soil salinity on quality and quantity forage yield of five alfalfa (Medicago sativa L.) varieties under Ahvaz conditions. Plant Production. 40(3); 113-127. Neupane, J., W. Guo, 2019. Agronomic basis and strategies for precision water management: A review. Agronomy, 9; 87-98. Pour-Farhad, A., F. Nurmandmoyed., S. Aharizadeh, A.A. Jafari, 2008. Grouping alfalfa ecotypes using multivariate statistical analyses. Scientific Research Journal of Agricultural Sciences, 3(9); 15-27. Rabiei, K., M. Khodam-Bashi, A. Rezaei, 2008. Identifying traits affecting potato yield using multivariate statistical methods under drought stress and non-stress conditions. Journal of Agricultural Science and Technology and Natural Resources, 46 (12): 23-37. 100 DESERT, 29-1, 2024 Ranjbar, A., A. Sepaskhah, S. Emadi, 2015. Relationships between wheat yield, yield components and physico-chemical properties of soil under rain- fed conditions. International Journal of Plant Production, 9; 433-466. Sharma, B.D., D.K. Hore, 1993. Multivariate analysis of divergence in upland rice. Indian, Journal of Agriculture Science, 63; 515-517. Solomon, S., D. Qin, M. Manning, K. Averyt, M. Marquis, M.M. Tignor, 2007. Climate Change 2007The Physical Science Basis: Working Group I Contribution to the Fourth Assessment Report of the IPCC, Cambridge University Press: Cambridge, UK, Volume 4. Tucak, M., S. Popović, T. Čupić, G. Šimić, R. Gantner, V. Meglič, 2009. Evaluation of alfalfa germplasm collection by multivariate analysis based on phenotypic traits. Romanian Agricultural Research, 26; 47-5. Zabat, M., A. Hosseinzade. 2011. Determining the most important traits affecting the yield of mung bean (Vigna radiate L. Wilczek) using multivariate statistical methods under non-stress and drought stress conditions. Iranian Legume Research Journal, 2 (1); 87-98.