Introduction

Agriculture has indeed played a crucial role in ensuring food security to feed the growing global human population1. Nevertheless, over-reliance on a handful of major staple crops reduced the diversification of our food basket which increased the risks of malnutrition, hunger and poor health. Although the underutilized crops have significant potential for contributing food and nutrition security, they are not widely grown or consumed2. Thus, exploring the potentiality of underutilized crops is essential to boost and diversify the agricultural production system in the face of climate change and biodiversity destruction3.

Among the various underutilized food crops, buckwheat (Fagopyrum spp.) is one such multipurpose crop owing to multiples uses like food, feed, medicine and manures4.Out of the about 30 species of the genus Fagopyrum, only two species namely Fagopyrum esculentum (common buckwheat) and Fagopyrum tataricum (Tartary buckwheat) are widely cultivated5. These two cultivated species harbor a significant amount of genetic diversity. Such genetic variation could be utilized for future crop improvement programs6. F. esculentum is mostly prevails in the temperate regions of the northern hemisphere, while, F. tataricum is predominantly a crop of higher altitudes7. In India, buckwheat is mostly cultivated along the Himalayan foothills from Ladakh in the North to Arunachal Pradesh in the East8. This crop is regarded as climate smart which can thrive in marginal and less input demanding condition as well9. Further, it is loaded with high-quality protein with balanced composition of essential amino acids, such as lysine, methionine and tryptophan which are the limiting amino acids in most cereal grains and legumes10. Buckwheat grains also contain various nutraceutical compounds and are rich in dietary fiber, macroelements (K, Ca, Na, Mg), microelements (Cu, Mn, Zn), and vitamins, particularly those of the B group, phenolic compounds including rutin, orientin, vitexin, quercetin, isovitexin, kaempferol-3-rutinoside, isoorientin, and catechins11. It is a valuable food supplements for regulating the blood sugar, blood lipid, blood pressure and helpful for individuals suffering from celiac disease due to its gluten free flour6,12,13. Thus, it is regarded as a crucial crop in mountainous regions across the northern hemisphere countries14.

Despite their immense potential and superiority, the lack of effective improvement measures compared to other crops has led to a noticeable decline in the global acreage of buckwheat species and an alarming depletion of their genetic diversity in recent years15.In view of the tremendous potential of this crop, focused breeding efforts are urgently required to capitalize on the genetic variation into enhancing production and productivity. Genetic diversity is the primary driver of selection and evolution and the genetic basis for breeders16. Hence, understanding genetic variation is crucial for comprehending the range of genetic variability and its possible application in breeding programs. To date, about 10,000 buckwheat accessions are stored in various gene banks around the world dedicated to the conservation of buckwheat genetic resources17. Research on the systematic evaluation of buckwheat genotypes using agro-morphological and molecular markers vis-à-vis the identification of superior parental lines is scanty, which makes the buckwheat improvement very difficult. Previous studies have also reported diversity of buckwheat using different molecular markers such as randomly amplified polymorphic DNA (RAPD)18, amplified fragment-length polymorphisms (AFLPs)19, inter-simple sequence repeats (ISSRs)20, and simple sequence repeats (SSRs)21,22,23. However, it provided insights into only a tiny fraction of the genetic diversity of common and tartary buckwheat24. Thus, genetic studies of diverse buckwheat gene pool and strategy to breed on local specific varieties in different regions of the world is necessary to ensure food and nutrition security25.

Further, the information on systemic characterization, promising donor lines and high yielding lines of buckwheat suited to north-eastern hill region of India is scanty and meager till date. Therefore, the current investigation was carried with the objectives to assess the morpho-molecular diversity and population differentiation of buckwheat germplasm of Fagopyrum esculentum and Fagopyrum tataricum, evaluate the performance of yield and component traits, and to identify promising high-yielding and stable genotypes with better nutritional properties for yield augmentation and as donor for future breeding program.

Materials and methods

Experimental location and plant materials

The study was conducted at Plant Breeding farm, ICAR-Research Complex for NEH Region, Umiam, Meghalaya, situated at a latitude of 25° 41’ 6.09” North and longitude of 91° 54’ 58.89” East and an altitude of 959 m above mean sea level. The soil of this region is acidic in nature with a pH of 5.2 to 5.8. It has high aluminium and iron and low phosphorous content. A set of 98 buckwheat genotypes were planted with four checks grown with a row-to-row spacing of 30 cm and a plant-to-plant spacing of 10 cm in an augmented design with 7 blocks and 4 checks in three environments, namely, Rabi−2022-23, Kharif-2023, and Rabi−2023-24.

Evaluation for agro-morphological and biochemical traits

The data for several biochemical characteristics, including total carbohydrates (%), protein (%), and phytic acid (%), as well as qualitative characteristics, including stem color, leaf margin color, leaf blade shape, blossom color, seed shape, and seed coat color, were collected from five randomly chosen plants. Plant height (PH), leaf blade length (LBL), leaf blade width (LBW), hundred seed weight (HSW), number of seeds per plant (NSPPT), petiole length (PETL), seed length (SEEDL), seed width (SEEDW), cyme length (CYMEL), and yield per plant (YPP) were among the quantitative observations. The suggested plant protection and agronomic practices were applied. The mean data of five plants was calculated and used for data analysis.

Molecular characterization

Characterization of the ninety-eight buckwheat accessions along with four checks was done at the molecular level through the use of 69 SSR markers. Young and healthy leaves from 15-day-old seedlings were collected and stored at −20 °C. The CTAB method26 with slight modifications was followed for DNA isolation. For quantification of DNA, the concentration was estimated using a Thermo Scientific Nano Drop 2000 Spectrophotometer. It measures the purity ratio (260/230 or 260/280) of DNA i.e., amount of RNA or protein contents by reading 260 absorbance. DNA samples were diluted with nuclease-free water accordingly from the obtained ratios. The genomic DNA was subjected to PCR amplification using a master mix containing 2 µl of genomic DNA, 1.2 µl each of forward and reverse primer, 0.2 µl of deoxynucleotide triphosphates (dNTPs), 0.2 µl of Taq polymerase, 1 µl of PCR buffer, and a final volume made of 10 µl by adding 4.2 µl of nuclease-free water. The PCR reaction was run for 35 cycles, and the PCR products were analyzed by gel electrophoresis using a 3.5% agarose gel. A 100 bp ladder was used to determine the base pair size of the amplified product. The amplified DNA fragments were then visualized under a UV transilluminator and documented using a gel documentation system (IGeneLabserve, Model IG-618GD). The alleles were assigned band sizes relative to the molecular size of the DNA ladder, and the missing bands were indicated as 0.

Statistical analysis

The quantitative data collected for 102 genotypes for different agro-morphological traits was used to examine genetic variability statistics, such as heritability, genetic advance over mean, and genotypic and phenotypic variances of the genotypes, as well as descriptive statistics, such as range, mean, standard error, and coefficient of variation using R 4.2.1 software27. All qualitative traits were simultaneously divided into discrete groups, and the frequency percentage values for the two species were determined. R 4.2.1 was also used to analyze the genotypic and phenotypic correlation coefficient, path coefficient (Variability package), and clustering (Stats package). The Factoextra tool was used to show the results. Additionally, PCA was carried out with R 4.2.1’s Factoshiny package. Data obtained for yield and contributing traits across three environments was subjected to AMMI analysis, a unified integration of analysis of variance (ANOVA) and principal component analysis (PCA)28,29, and stable genotypes in the experimental hybrids across the testing environments were identified; the significance of which was tested by F-test procedure30. AMMI biplots were constructed based on genotype main effect means versus the first interaction principal component axis (IPCA1) and also between first two interaction principal component axes (IPCA1 vs. IPCA2)31. Various AMMI-based stability measures have been utilized in the present study to describe the yield stability of buckwheat genotypes. These AMMI-based stability measures were calculated using the package Metan of R Studio 4.2.1.

Different AMMI Stability measures used in the study are-

  • Modified AMMI Stability Index (MASI) =\(\:\sqrt{{\sum\:}_{n=1}^{N{\prime\:}}\left({PC}_{N}^{2}X{\theta\:}_{n}^{2}\right)}\)[15]

  • Modified AMMI Stability Value (MASV) =\(\:\sqrt{{\sum\:}_{n=1}^{{N}^{{\prime\:}}-1}\left(\frac{{SSIPC}_{n}}{{SSIPC}_{n+1}}\right)x{\left({PC}_{n}\right)}^{2}+{\left({PC}_{N"}\right)}^{2}}\)[15]

  • Sums of the Absolute Value of the IPC Scores (SIPC) =\(\:\sum\:_{n=1}^{N{\prime\:}}\left|{{\lambda\:}_{n}}^{0.5}{\gamma\:}_{in}\right|\)[16]

  • Absolute Value of the Relative Contribution of IPCs to the Interaction (ZA) =\(\:\sum\:_{n=1}^{N{\prime\:}}\left|{\theta\:}_{n}{\gamma\:}_{in}\right|\)[17]

The multi-trait stability index (MTSI), was also calculated in metan package of R 4.2.1 using the equation as follows:

$${\rm MTSI_{i}} =\:{\left[{\sum\:}_{j=1}^{f}{\left(Fij-Fj\right)}^{2}\right]}^{0.5}$$

Where, MTSI is the multi-trait stability index for the ith genotype, Fij is the jth score of the ith genotype, and Fj is the jth score of ideotype. The genotype with the lowest MTSI is, therefore, closer to the ideotype and hence has a high mean performance and stability for all variables studied.

Biochemical analysis

For biochemical investigation, extreme genotypes from both buckwheat species were chosen based on yield. The Anthrone reagent method32, was used to estimate the amount of carbohydrates. Lowry’s method (1951) was used to assess the protein content33, while Wheeler and Ferrel’s (1971) approach34 was used to quantify the phytic acid, an anti-nutritional component.

Marker data analysis

Using Power Marker 3.2535 and Popgene 1.32 software36, the data was analyzed for allele frequency, gene diversity, observed heterozygosity, Shannon’s information index, and polymorphic information content (PIC). Genetic diversity was calculated as per Nei’s (1973) genetic distance using the unweighted pair-group arithmetic average method (UPGMA) in Power Marker 3.25, and Mega 11.0 was used for visualization. The first two principal coordinates of principal coordinates analysis (PCoA) were calculated using GenAlEx 6.537. The model-based genetic structure of 104 buckwheat accessions was analyzed using the STRUCTURE software38. Bayesian cluster estimation was performed with ten replicates for each K value (1 to 10). The program inferred population structure and assigned individuals to populations based on SSR genotypes, using a burn-in and run length of 100,000, allowing for admixture and correlated allele frequencies. The likelihood of the data for various K values was calculated, with probabilities becoming similar for higher K values. The optimal range of K values was determined by genotypes and marginal likelihood to explain the data structure using the Structure Harvester program39.

Results

Evaluation of buckwheat germplasm for agro-morphological traits

Days to 50% flowering ranged from 29.0 (IC16558) to 57.0 (IC13144) with an average of 47.1 (Table 1). Likewise, days to 80% maturity also ranged from 76.0 (IC26594) to 103.0 (EC18751) with an average of 92.23. The maximum number of days to 80% maturity was 103 days, while the minimum number of days to 80% maturity was 76 days (IC17370). Plant height ranged from 20.33 to 89.67 cm with an average of 48.77 cm. IC26592 (89.67 cm) had maximum plant height, followed by IC26586 (82 cm) and IC37287 (79 cm), while shorter plant height was observed in IC107960 (20.33 cm) and IC107962 (20.33 cm), followed by IC107989 (21 cm). Cyme length also varied among genotypes from 2.33 (IC107989) to 17.17 cm (IC107021) with an average of 7.55 cm. The length of leaves among genotypes varied from 2.73 cm to 11.60 cm with an average of 6.10 cm. IC37307 (11.60 cm) had maximum leaf length, followed by IC37297 (10.37 cm) and IC37302 (10.10 cm), while IC107975 (2.73 cm) and IC107021 (2.80 cm) had minimum leaf length. Likewise, leaf width among genotypes varied from 1.73 (IC107975) to 9.30 cm (IC37297) with an average of 4.58 cm. The seed width ranged from 1.13 cm to 5.70 cm with an average of 2.92 cm. EC18751 (5.70 cm) exhibited maximum seed width followed by IC37265 (4.93 cm) and IC24296 (4.93 cm), while minimum seed width was observed in IC16550 (1.13 cm) and IC49680 (1.23 cm), respectively. The length of the seed ranged from 2.93 (IC26586) to 9.17 cm (EC18040) with an average of 5.58. As far as petiole length is concerned, it ranged from 1.23 (IC42423) to 12.17 cm (IC37305) with an average of 5.62. Hundred seed weight among genotypes varied from 1.00 to 3.50 with an average of 2.13. Maximum 100 seed weight was found in IC107971 (3.50 g), followed by IC37293 (3.2 g) and IC16551 (3.2 g), whereas minimum 100 seed weight was found in IC13145 (1.00 g) and IC49681 (1.1 g). Likewise, the number of seeds per plant ranged from 9.67 to 424.67 with an overall mean of 110.47. The maximum number of seeds per plant was observed in IC16550 (424.67), followed by Himpriya (412.00) and IC26586 (366), while the minimum number of seeds was observed in IC107971 (9.67), IC107274 (11.33), and IC107583 (12.33). As far as yield per plant is concerned, it varied from 0.20 to 10.19 g with an average of 2.35 g. Maximum yield per plant was found in IC16550 (10.19 g), followed by IC26595 (9.54 g) and IC26586 (8.42 g), while minimum yield per plant was found in IC107583 (0.20 g), IC107274 (0.24 g), and IC107807 (0.24 g).

Table 1 Genetic parameters of variation in buckwheat genotypes.
Full size table

Evaluation based on qualitative traits

Two types of stem color were observed in the current set of genotypes, i.e., pink and green (Fig. 1). The genotypes with green stem color were identified as tartaricum species, e.g., IC13412, IC16550, IC16551, Shimla B-1, Local Teethey, Himpriya, etc., and those with pink color stems were identified as esculentum species, e.g., IC14890, IC16554, IC16558, IC16559, and Local Meethey. Similarly, two types of flower color were observed, based on which genotypes were identified as tartaricum and esculentum species. The genotypes with greenish-yellow flower color were identified as tartarium buckwheat, e.g., IC16556, IC16557, EC18673, IC17371, etc., and those with pink flower color were the esculentum species, e.g., IC26594, IC37265, IC37266, IC37269, etc. Pink and green leaf margin color was found among the genotypes. Pink leaf margin color was found in IC13412, IC14890, IC16550, Local Meethey, Himpriya, etc., and green leaf margin color was found in EC18049, EC18726, IC24298, IC24302, IC26586, etc. Four different types of leaf shapes were also observed among the genotypes, which were hastate, cordate, ovate, and sagittate. IC13412, IC14890, IC16556, EC18673, etc., showed a hastate type of leaf shape; IC16550, IC16551, IC13144, IC13145, etc., genotypes had a cordate type of leaf shape; IC16558, EC18040, and IC24302 had an ovate leaf shape, whereas IC107807, IC107631, IC107274, IC107975, and IC107976 had a sagittate type of leaf shape. Three different types of seed shapes, namely conoidal, ovate, and triangular, were found in the genotypes. Conoidal seed shape was found in IC107989, local teethey, IC107974, IC107967, IC107960, etc., whereas ovate seed shape was found in IC14890, IC13144, IC13145, IC13411, etc., and triangular seed shape was found in EC18653, IC37265, IC37287, IC37293, IC37311, IC107021, etc. In the same manner, four different types of colors were present in the seed coat of the genotypes, i.e., brown, mottled, grey, and black. IC13411, IC26586, IC49677, IC13413, IC107970 etc. had brown seed colour, Shimla B-1, Himpriya, IC42407, EC18751 etc. showed grey seed color and IC26592, IC107987, IC107564 etc. were found to have black seed colour whereas Local meethey, EC18653, IC79192, IC107807 have shown mottled colour. The maximum number of tartaricum genotypes showed a grey color of seed coat, whereas for esculentum species, it was a mottled color.

Fig. 1
figure 1

Morphological variation in buckwheat accessions.

Full size image

Assessment for genetic variability parameters

The GCV and PCV parameters give us a better idea about character heritability and the extent to which the trait is influenced by environment. The highest phenotypic coefficient of variation was observed for yield per plant (74.12), followed by number of seeds per plant (70.66) and petiole length (35.66). In the same manner, yield per plant (70.30) recorded the highest genotypic coefficient of variation, followed by the number of seeds per plant (68.31) and petiole length (34.42). For traits like seed width, 100 seed weight, seed length, leaf blade length, leaf blade width, cyme length, and plant height at maturity, the PCV and GCV values are statistically at par, showing very little influence of the environment over the trait. A wide range of heritability was also observed among the traits. Plant height (96.87) showed the highest heritability, followed by the number of seeds per plant (93.44). Among all traits, days to 50% flowering (88.42), petiole length (98.34), yield per plant (89.97), cyme length (82.33), seed width (92.44), seed length (79.12), 100 seed weight (92.33), leaf width (70.14), and days to 80% maturity (63.04) showed high heritability, whereas leaf length (54.91) had medium heritability (Table 1). Genetic advance as a percentage of the mean was analyzed to predict the genetic gain of the trait over the mean. The highest genetic advance was recorded for the number of seeds per plant, followed by yield per plant and petiole length.

Correlation studies for yield and related traits

Pearson’s Correlation Coefficient analysis was performed to examine the relationship among various traits. This analysis helps breeders enhance grain yield by selecting traits that contribute to higher yields (Table 2). In the correlation study, yield per plant showed a positive significant correlation with cyme length, hundred seed weight, days to 50% flowering, and days to 80% maturity and also showed a negative significant correlation with seed length and seed width. Plant height at maturity was significantly and positively correlated with the number of seeds per plant, yield per plant, petiole length, cyme length, days to 50% flowering, and days to 80% maturity. Likewise, leaf blade length showed a positive correlation with leaf blade width and petiole length and showed a negative significant correlation with cyme length. Seed length showed significant positive correlation with seed width and negative correlation with cyme length. The number of seeds per plant showed significant positive and high correlation with yield per plant, followed by cyme length, days to 50% flowering, and days to 80% maturity, and showed significant negative correlation with seed length and seed width.

Table 2 Phenotypic and genotypic correlation coefficients among yield and attributing traits.
Full size table

Path coefficient analysis

Path coefficient analysis divides the correlation coefficients and provides insights into the direct and indirect influences of multiple traits on yield. Among all the traits studied, the number of seeds per plant had the highest positive direct effect (0.9161) towards yield, followed by the hundred seed weight (0.3094). Positive direct effects were also recorded for petiole length, leaf blade width, seed length, and days to 80% maturity, while negative direct effects were observed for leaf blade length, plant height, days to 50% flowering, seed width, and cyme length (Table 3). As far as indirect effects are concerned, days to 50% flowering showed a positive indirect effect on seed yield per plant through plant height, petiole length, cyme length, 100 seed weight, and seed length. Plant height at maturity showed positive indirect effect on seed yield per plant through days to 80% maturity, days to 50% flowering, seed width, seed length, 100 seed weight and Leaf blade width. Leaf blade width showed a positive indirect effect on seed yield per plant through leaf blade length, petiole length, seed length, seed width, days to 50% flowering, and days to 80% maturity. Petiole length showed a positive indirect effect on seed yield per plant through plant height, leaf blade length, leaf blade width, number of seeds per plant, seed length, seed width, and cyme length. 100 seed weight showed positive indirect effect on seed yield per plant through Leaf blade length, Seed length, Seed width, cyme length, days to 50% flowering and days to 80% maturity.

Table 3 Path coefficient analysis showing direct and indirect effect of observed traits on yield.
Full size table

Cluster analysis based on quantitative traits

Based on cluster analysis using euclidean distance, the buckwheat accessions were grouped into 5 major clusters (Fig. 2). Cluster V is the largest group (24 genotypes), while cluster I (17 genotypes) was the smallest among the five clusters. As seed yield is one of the most important traits for the breeder, it is observed that cluster V included germplasms with maximum mean value for yield and yield-contributing traits like plant height, leaf blade length, leaf blade width, number of seeds per plant, yield per plant, and petiole length (Supplementary Tables 1 and 2). The highest average number of seeds per plant was recorded in cluster V (161.01), followed by cluster III (124.77), and the least in cluster I (43.73), while the maximum mean 100 seed weight was recorded in cluster III (2.37), followed by cluster II (2.19), and the least in cluster I (1.92). Cluster V (3.36) had the maximum average seed yield per plant, followed by cluster III (2.71) and the least in cluster I (0.83). Genotypes of cluster IV have taken maximum time (50.81) for flowering, whereas the minimum was recorded for cluster II (39.24). The highest average plant height was observed in cluster V (59.93), followed by cluster II (59.68) and the minimum were observed in cluster I (32.67). Mean seed length was recorded maximum in cluster II (5.99), followed by cluster V (5.84) and least in cluster III (5.00), while mean seed width was recorded maximum in cluster II (3.41), followed by cluster I (3.10) and least in cluster IV (2.61).

Fig. 2
figure 2

Cluster diagram of buckwheat genotypes based on Euclidean distance.

Full size image

Principal component analysis (PCA)

The dimension of data was reduced through principal component analysis to enhance data interpretation efficiency while retaining as much of the original information from the observations as possible. PCA of morphological and yield-related traits showed that 61.05% of total variation was accounted for by the first three principal components. The first, second, third, fourth, and fifth principal components accounted for about 23.53%, 20.71%, 16.81%, 10.34%, and 8.33% of the total variation, respectively (Table 4). The number of seeds per plant and yield per plant were the major contributing traits for the first principal component (PC1). Similarly, plant height and petiole length were the major contributing traits for the second principal component (PC2), whereas in the third principal component (PC3), leaf blade length and leaf blade width were the major contributors. IC37307, IC37292, IC16550, IC26586, IC26595, IC16550, IC37275, IC107989, IC107976, IC107977, and IC107989 were the genotypes present at extreme points of the PCA individuals biplot (Fig. 3). The expected outcome is for these lines to exhibit greater genetic diversity and variability in the traits that contribute towards yield.

Table 4 Eigen values and variance percentage of various dimensions of the PCA.
Full size table
Fig. 3
figure 3

Genotype by trait (GT) Biplot for the buckwheat genotypes for agro-morphological traits.

Full size image

Stability analysis by AMMI model

Based on AMMI Analysis of Variance for agro-morphological traits in buckwheat genotypes across three environments, the environments, genotypes, and GEI (genotype x environment interaction) showed significant variation for the twelve agro-morphological traits under study. The GEI was further decomposed into two principal component axes (IPCAs), accounting for 100% of the total variation for all the traits (Table 5). The two PCAs fitted in the current study were also found to be significant (p < 0.001) for all the twelve agro-morphological traits.

Table 5 AMMI ANOVA (Analysis of Variance) for the buckwheat genotypes studied over three environments.
Full size table

AMMI biplot

The phenotypic performance and stability of 102 buckwheat genotypes under three different environments were graphically visualized using GEI biplots. For yield per plant, the cumulative percentage of the GxE interaction that was justified by PC1 and PC2 was 100%. Also, the contributions of PC1 and PC2 were 83.9% and 16.1%, respectively. Among the three seasons, season 3 had the highest average yield, and genotypes IC16550, IC42424, IC26606, EC18049, and Shimla B-1 had shown higher and more stable yielding behavior than the overall mean yield (Fig. 4). For the number of seeds per plant, the cumulative percentage of the GxE interaction (100%) was contributed by PC1 and PC2 in the range of 80.8% and 19.2%, respectively (Supplementary Fig. 1). The genotypes Himpriya, IC37275, and IC26592 showed the highest number of seeds per plant and showed stability towards showing the trait. To determine the genotype consistency across three environments, a number of AMMI-based stability measures have been computed. A better genotype yield consistency is indicated by a lower stability measure value. Based on different AMMI stability parameters such as MASI, ZA, MASV, and SIPC genotypes, IC37280, IC17371, IC26594, IC13139, IC42416, IC107608, IC16557, IC107967, EC18751, IC13411, and IC24302 were found to be the most stable ones with respect to yield per plant. Among the genotypes, IC13139, IC37280, and EC18751 were found to be good yielders also based on mean performances and thus are well adapted to the variable environmental conditions with higher yield potential.

Fig. 4
figure 4

AMMI biplot of 102 genotypes for yield per plant across three environments.

Full size image

Multi-trait stability index (MTSI)

The MTSI was computed using the factor components retained by an exploratory factor analysis (EFA), and the accumulated variance contributed by these components was 68.51%. The mean commonality was 0.694 with a range of 0.402 (for HSW) to 0.922 (for NSPPT). The twelve traits studied were grouped into five factors (FA) by extracting WAASBY scores from each character. According to this, NSPPT and YPP fall under FA1; DFF, PH, and DM fall under FA2; HSW and LBW fall under FA3; and LBL and PETL fall under FA4. SEEDL, SEEDW, and CYMEL fall under FA5 since they have the major contribution towards it (Table 6). Selection differential (SD) for both the mean performance and WAASBY index for 12 traits were also estimated. Maximum traits had a positive selection differential (SD) for the WAASBY index, referring to the method’s effectiveness in identifying the most stable and productive genotypes. The mean SD% for the WAASBY index was 10.31%, with the lowest one (−8.1%) for the SEEDL and the highest for NSPPT (26.4%). A negative SD% for mean performance of the traits viz., HSW, SEEDL, and SEEDW was also observed. Assuming a selection intensity of 20%, the fifteen genotypes, viz., IC37275, IC26592, local teethey, Shimla B-1, IC13139, IC36919, IC26606, IC26586, IC107987, IC107974, IC13412, IC13145, IC37269, IC107608, and IC26595, were selected (Fig. 5).

Table 6 Explained variance, factorial loadings after varimax rotation and communalities obtained in the factor analysis for 12 traits of 102 buckwheat germplasm evaluated for three seasons.
Full size table
Fig. 5
figure 5

MTSI plot showing genotype ranking and selected genotypes.

Full size image

Molecular characterization of buckwheat genotypes using SSR markers

In this work, out of 69 SSR primers were used to evaluate polymorphism and amplification across 102 genotypes, 65 (94.2%) amplified well and 41 (59.42%) showed polymorphism. For the polymorphic SSR markers, the PCR-amplified products varied in size from 75 bp to 290 bp. Table 7 displays the statistical summary of genetic variation for each SSR locus. A total of 172 alleles, or 2.65 alleles per locus, were found using 65 markers. The number of alleles ranged from 2.00 to 4.00, whereas the effective number of alleles ranged from 1.10 to 2.78 with an average of 1.62. With an average of 0.616, the major allele frequency varied between 0.446 and 0.814. The two markers with the highest major allele frequencies were Fes1693 (0.814) and Fes1322 (0.784), closely followed by Fes3177 (0.765). On the other hand, Fes3164 (0.446) and Fem2751 (0.451) had the lowest major allele frequencies. With an average of 0.082, the observed heterozygosity for the polymorphic markers ranged from 0.020 (Fem1407) to 0.363 (GB-FE-080). With an average of 0.409, the polymorphic information content (PIC) values varied from Fes1693 (0.257) to Fes3164 (0.564).

Table 7 The summary of genetic variation of 41 polymorphic SSR markers.
Full size table

Cluster analysis based on Nei’s genetic distances grouped the buckwheat accessions into 5 clusters. (Fig. 6). Cluster I included genotypes from 15 different locations, among which three were from East Sikkim. Cluster II included genotypes from 13 different locations, among which three were from Kinnaur. Both the clusters comprised of maximum genotypes from esculentum species. Cluster III included genotypes from 12 different locations, and the maximum genotypes were from Kullu and Kinnaur. Cluster IV included genotypes from 15 different locations; among them, the maximum was from Mandi and Kullu. Cluster V included genotypes from 8 different locations. An AMOVA was also conducted to quantify the variation among and within the subpopulation of 22 different origins. It was inferred that 84% of existing variation was due to among individuals within the population, whereas 14.44% was mainly due to variation within individuals (Table 8). The high percentage of variation showed the diverse nature of genotypes within a population. The genetic diversity and population structure of all the 98 buckwheat genotypes, along with four checks, were revalidated with principal coordinate analysis (PCOA). The first principal coordinate explained 17.75% of the variation among the accessions, while the second and third principal coordinates explained 4.18% and 3.94%, respectively. All three components together explained 25.87% of total variation (Supplementary Table 3). The genotypes were divided among two major groups (Fig. 7). Almost all the genotypes were distributed between two species i.e. esculentum and tartaricum irrespective of their location indicating its diversity and shared genetic relationship with each other. Genotypes like IC16558, EC18653, IC37307, and IC107965 were found to be present at the middle of two clusters, which indicates the mixing of genes in these lines.

Fig. 6
figure 6

Cluster diagram of genotypes based on Nei genetic distance.

Full size image
Table 8 Analysis of molecular variance (AMOVA).
Full size table
Fig. 7
figure 7

Genetic Diversity among buckwheat germplasm based on Principal coordinate analysis (PCoA).

Full size image

Population structure analysis based on a model-based clustering approach with various values of K ranging from 1 to 10 was performed to describe the population structure of buckwheat genotypes of different origins. Based on increasing likelihood ratios using delta K statistics, the ideal K value at K = 2 was selected based on the delta K value (Supplementary Table 4) with the highest peak (Supplementary Fig. 2), which divided the genotypes into two subpopulations. Further, genotypes present in each subpopulation were categorized as pure and admixtures by taking the membership fraction as 0.80. A high level of admixture with less than 80% from the estimated ancestry was present in the F. esculentum groupas compared to the F. esculentum group. Subpopulation I included the highest number of germplasms along with a high number of admixtures (Fig. 8).

Fig. 8
figure 8

Population structure diagram of 102 buckwheat genotypes based on estimate of Delta K = 2.

Full size image

Biochemical characterization of buckwheat genotypes

The data for all the biochemical parameters under study (protein, carbohydrate, and phytic acid) for selected genotypes were subjected to ANOVA analysis to test the significant difference among genotypes. Significant variations at P < 0.01 among the evaluated genotypes were observed for all the traits (Table 9 and Supplementary Table 5). Total protein significantly varied among the selected genotypes with an overall mean of 8.48%. Maximum total protein was recorded (Supplementary Table 6) in local teethey (11.45%), followed by IC37309 (10.55%), and the lowest was recorded in IC107966 (5.88%). Likewise, total carbohydrate content varied significantly among the genotypes (72.5–55.5%) with an average of 63.27%. The highest carbohydrate percentage was observed in IC107963 (72.5%), followed by IC37309 (69.6%) and IC107274 (68%). Phytic acid content also varied significantly among the genotypes (1.23 to 0.48 g/100 g) with an average of 0.837 g/100 g. The highest phytic acid content was observed in Himpriya (1.23 g/100 g), followed by IC37275 (1.22 g/100 g) and IC26586 (1.18 g/100 g). The lowest phytic acid content was observed in IC107991 (0.48 g/100 g). All three nutritional parameters were found to be higher in the T. tartaricum species in comparison to the T. esculentum species. A high coefficient of variation was found in the case of phytic acid as compared to the other two biochemical traits, which indicates that the trait is more variable. All three biochemical traits had a high positive significant correlation with each other. Also, these biochemical traits had a positive significant correlation with traits like days to 80% maturity and days to 50% flowering but showed a positive non-significant correlation with the yield per plant.

Table 9 Biochemical parameters of variation in buckwheat.
Full size table

Discussion

Consistent over-reliance on a handful of major staple crops increased the risks of hidden hunger. In this regard, some of the conventional food crops like buckwheat with promising nutritional values must be mainstreamed to diversify our food basket. Thus, exploring the genetic diversity and population structure in buckwheat germplasm is imperative to mine desirable genes from elite genotypes40, and to augment their utility in breeding programs for ensuring food and nutritional security41.

Our results indicates that the analyzed set comprising of tartaricum and esculentum species exhibits great variation in terms of agro-morphological features as well as at molecular level. The study found interesting correlations and relationships among the studied parameters. The discriminating technologies of clustering and PCA analysis also enabled in visualization of underlying relationships among studied buckwheat accessions. This indicates that high-quality buckwheat accessions could be selected from the analyzed set. Thus, most promising accessions with high yielding and stable across environments with desirable nutritive value have been identified from the analyzed set of 102 buckwheat genotypes.

In terms of the quantitative traits diversity, our study advocates presence of substantial variability such as days to 50% flowering varied from 29.0 to 57.0, days to 80% (76.0 to 103.0), plant height (20.33 to 89.67 cm), cyme length (2.33 to 17.17 cm), leaf length (2.73 cm to 11.60 cm), seed width (1.13 cm to 5.70 cm), seed length (2.93 to 9.17 cm), test weight of 100 seed (1.00 to 3.50), number of seeds per plant (9.67 to 424.67), and yield per plant (0.20 to 10.19 g) respectively. As per yield per plant, tartary buckwheat was found to be a higher yielder than the common buckwheat. The genotypes differed significantly from one another set for yield and qualities. High yielders among the genotypes were IC16550 from Kullu, IC37275 from Namchi, and IC26595 from Lahaul and Spiti. For traits like yield per plant, 100 seed weight, number of seeds per plant, and petiole length, both PCV and GCV values were on a higher node. The difference between genotypic and phenotypic coefficients of variation, as observed in our experiment, was merely a reflection of the environment on the expression of traits. The observations for the agro-morphological traits variation were found to be in agreement with previous studies24,42,43. Further, the estimations of genetic advancement rely on another parameter, namely, heritability. A higher heritability estimate for a trait suggests that environmental factors have less impact on its expression, making it suitable for improvement through simple selection methods. In the present study, the highest heritability was observed in plant height, followed by the number of seeds per plant. In the same manner, the genetic advance as a percentage of the mean was also observed to be higher for yield per plant, number of seeds per plant, and petiole length. As heritability combined with genetic advance over the mean offers a trustworthy estimate of both additive and non-additive effects, future breeding programs may utilize these attributes as the foundation for selecting superior breeding lines44. The presence of additive gene effects in the inheritance of these traits eventually makes selection for these features extremely successful. In any breeding programs and documentation, understanding the description based on phenotypic characters is prominence45. The observations revealed by studies indicated that common buckwheat germplasm poses considerable diversity in morphological traits41. One possible reason could be due to outcrossing, mediated via insect cross-pollination46,47.

On the basis of qualitative characters, namely flower color, stem color, and seed coat color, the set of genotypes was clearly divided into two species, i.e., esculentum and tartaricum. It has been reported that the two species differ greatly in their growing and adaptation to different environments. The stem color of the genotypes was green in tartaricum and pink for esculentum species respectively. Similarly, the flower color of tartarium species were greenish-yellow and pink flower color predominate in the esculentum species. The maximum number of tartaricum genotypes showed a grey color of seed coat, whereas for esculentum species, it was a mottled color. Further, the leaf shape varied widely such as hastate, cordate, ovate, and sagittate among the genotypes. Likewise, seed shapes varied from conoidal, ovate to triangular shapes. In consistent with our finding of flower color variation, the prevalence of greenish-yellow, white and pink flower color among two species of buckwheat have also been reported41,48. In buckwheat, the color of plant vegetative organs as well as their flavonoid content is affected by different factors such as variety and environmental conditions49,50. In our study, the prominent stem color was green and pink. The amount of rutin content, a significant flavonoid, may be estimated by the stem color47. The stem color variation (red and pink) in buckwheat germplasm could also be attributed to nature of anthocyanin accumulation in different environment51,52. In line with our study, wide variations in seed shape and seed color have also been previously reported in north-west Indian buckwheat landraces53. The prevalence of conoidal, ovate and triangular seed shapes in our study also in agreement with early report wherein, ovate and triangular predominate among the buckwheat germplasm54.

Understanding the correlation between traits helps in identifying a specific trait that can impact seed yield, as it is a complex quantitative trait; direct selection for this trait may not be very profitable. Thus, correlation between key traits can serve as a useful indicator for the genetic improvement41. Our study revealed that yield per plant exhibited a positive significant correlation with days to 80% maturity, days to 50% flowering, and cyme length. Test weight also showed a positive correlation with yield per plant, while the association is negatively significant with seed length and seed width. The positive correlation between yield per plant with days to flowering, days to maturity and hundred seed weight in our study is consistent with earlier study42. The results of25 also confirm our findings. In addition, the regression-based path analysis, a statistical tool that allows for the examination of the direct and indirect influence of different traits. According to numerous other researchers55, path analysis provides a comprehensive representation of the causal variables that influence the final outcome of the research, which in this case is the seed yield per plant. The number of seeds per plant, followed by the hundred seed weight, had the most positive direct influence on yield among all the variables examined. In other experiments it was also observed that through path analysis that days to maturity, leaf length, leaf width, and 100 seed weight had a direct positive effect on yield per plant. This is congruence with earlier report42. The findings of56 were also similar to our findings to some extent. This implies that the trait in concerned were more of additive gene action. Thus, selection and improvement for such prominence traits can eventually increase yield per plant.

The cluster analysis and PCA helped in grouping the accessions, effective management and utilization of germplasm4. In the present investigation, the morphological clustering based on Euclidean distance of 12 quantitative traits grouped the analyzed set into 5 major clusters. Since one of the most crucial characteristics for a breeder is seed yield, accessions in cluster V had maximum cluster mean of yield and companion traits such as petiole length, plant height, leaf blade length, and width, number of seeds per plant and yield per plant. Likewise, accessions in cluster-II exhibit shortest cluster mean of days to flowering (39.24) and days to physiological maturity (87.4). Such genetic differentiation and visualization would be of immense value for breeding for specific traits and parental selection program. Similar grouping of local buckwheat germplasm into 5 clusters using quantitative characters had also been reported57. Further, previous study also reported wide morphological variation and grouping of common buckwheat (Fagopyrum esculentum) and tartary buckwheat (Fagopyrum tataricum) germplasm of Northwest Indian Himalayas41. The diversity observed in our study was also similar to earlier studies under field conditions for common and Tartary buckwheat collections from China58, Nepal48, India41,51,59 and Japan60.

In PC analysis, the degree of trait contributions in the first three principal components is crucial61. In our study, the first three principal components of the PCA of morphological yield-related traits accounted for 61.05% of the overall variation. This suggest that the trait included in the study largely determined the total genetic variation of 102 buckwheat accessions. Based on the trait loading in the first two PCs, the major contributors in the first principal component (PC1) were the number of seeds per plant and the yield per plant. In PC2, trait like plant height and petiole length exhibit strong positive loading, while, in the third main component (PC3), the most contributing traits were leaf blade length and leaf blade width. The first three components in another study reported 59.97% and 74.36% variance in common and tataricum type buckwheat, respectively53. Thus, in addition to clustering, PCA is effective in comprehensive crop evaluation, and valuable scientific foundation for parent selection in breeding programs62. Hence, our results of agro-morphological characterization indicate that the buckwheat accessions analyzed in our study harbor significant phenotypic diversity with scope of parental selection for future breeding programs.

Further, AMMI analysis of variance for the twelve agro-morphological traits in the present investigation showed that genotypes, environments, and GEI (genotype x environment interaction) showed substantial and significant variation. Additionally, the results based on the decomposition of the GEI variable into two principal component axes (IPCAs) were found to be significant (p < 0.001), accounting for 100% of the total variation for all the traits. According to the AMMI biplot for yield per plant, IC16550 was found to be the greatest yielder and most stable in seasons one and three. IC13139, IC37280, and EC18751 were also determined to be stable genotypes and good yielders based on a number of AMMI stability parameters. They can therefore produce good yield and are well suited to the three test environments. Further, the findings of the multi-trait stability index, including all the traits together, showed that the majority of attributes exhibited positive selection differentials (SD) for the WAASBY index, indicating that the technique was effective in identifying the stable and high-performing genotypes. The fifteen genotypes (IC37275, IC26592, Local teethey Shimla B-1, IC13139, IC36919, IC26606, IC26586, IC107987, IC107974, IC13412, IC13145, IC37269, IC107608, and IC26595) were chosen, assuming 20% selection intensity. Genotype performance is greatly influenced by Genotype × Environment Interactions (GEIs), which complicate the selection of stable, high-performing genotypes63. Thus, understanding of GEI pattern and yield stability analysis are crucial for development of high yielding and widely adaptable cultivars64. The use of AMMI analysis helps for the assessment of the overall GEI effect of each genotype, as well as their individual contributions65. To overcome the selection of stable genotypes but low yield, a mixed-effect model-based superiority index, WAASBY was developed by66. In addition, multi-trait stability index, including all the traits efficiently and simultaneously selects stability with high mean performance, taking into account various attributes with different weights67. Thus, the varietal recommendation based on mean performance and stability of multiple agronomically desirable traits would be more reliable68.

Molecular characterization of buckwheat genotypes using SSR markers

Microsatellite markers (SSRs) are highly polymorphic, stable, and co-dominant in nature. It has been used for diversity and genetic relationship analysis of buckwheat46,69 and thus are effective tool for analyzing the genetic diversity and population structure of buckwheat70. In this study, out of 69 SSR primers used to evaluate polymorphism and amplification across 102 genotypes, a total of 172 alleles were produced, with an average of 2.65 alleles per locus. The values of the effective number of alleles were lower than the total number of alleles observed in the population. This might be due to the unequal distribution of gene frequencies across the SSR loci. With an average of 0.409, the polymorphic information content (PIC) values varied from 0.257 to 0.564. Fes2900, Fes1497, Fes2857, Fes2798, Fes2976, Fes3214, GB-FE-035, and Fes3164 were identified as the most discriminative markers with a PIC value of more than 0.5. Similar report on the genetic diversity in buckwheat was also using SSR markers has been reported [Song et al. 202258. Major allele frequencies (MAF) for the 10 polymorphic loci varied from 0.12 to 0.97 with a mean allele frequency of 0.57. Polymorphic information content (PIC) value varied from 0.257 to 0.564 with an average of 0.409 in the present study. This was in agreement with the studies conducted by58,71. Likewise, gene diversity (0.491), and heterozygosity (0.082) observed in our study also in fair agreement with previous studies on buckwheat46,71. Diversity analysis based on Nei’s genetic distances grouped the buckwheat accessions into 5 clusters. Mostly the clusters were not location-specific in the present study. Genotypes from F. esculentum species were mostly found in cluster I and cluster II. Fifteen accessions of F. tataricum were divided into two clusters (cluster 1 and cluster 2). The grouping of buckwheat accessions from different geographical regions, however, showed no region-based clustering and instead a complicated distribution pattern. This implies that because of strong migration pressure into new regions, buckwheat is broadly distributed with little local differentiation. In another study, 179 accessions were divided into 3 groups/clusters by dendrogram construction based on Nei’s genetic distance [Song et al. 202258 and observed the same. Earlier studies in various other crops also reported a low correlation between molecular grouping and geographical place of collection of accessions72. To measure the variation within and among 22 subpopulations of various origins, an analysis of molecular variance (AMOVA) was also carried out, revealing that 84% of the existing variation came from among individuals within the population, 14.44% from within individuals among the population, and 1.56% from among populations. Similarly, a total variation of 83.49% and 16.5% attributed to genetic diversity among, and within the population was reported71. The self-incompatible nature of F. esculentum may be the cause of the higher molecular variance seen among individuals within the population73. It was found that while variations among accessions accounted for 78.09% of the genetic variance, there was a notable 21.09% variation between the two species and no variation between accessions within the same species24. The variation among the accessions was assessed by principal coordinate analysis (PCoA), revealing that the first principal coordinate explained 17.75%, whereas the second and third principal coordinates explained 4.18% and 3.94%, respectively. Together, the three coordinates accounted for 25.87% of the variation in the whole. The differentiation between the two genetically distinct groups of F. esculentum and F. tataricum accessions was evident from the biplot showing that they were diverse and shared genetic components. The findings supported the results24, who found that the first two PCoA components explained 19.52% and 6.23% of the genetic variance, respectively. The first two major coordinates in the analysis46 for the SSR data set explained 32% of the information in the genetic identity matrix, which was more than in the current investigation. A model-based approach was used to estimate the population structure, showing that the delta K value peaked at K = 2, dividing all buckwheat germplasms into two subpopulations and suggesting the existence of two genetically different clusters that separate F. esculentum from F. tataricum. A larger number of admixtures were present in F. esculentum when compared to F. tataricum. This can be understood by the phenomenon of self-incompatibility. 78 genotypes were reported to be admixtures with mixing of genes from ancestry in a smaller to greater extent. Based on 10 SSRs, a population structure was created [Song et al. 202258 for 179 buckwheat accessions and discovered two subpopulations, each of which had 65 admixtures, in agreement with the current results. Different studies also showed that the SSR markers effectively differentiate between the Fagopyrum esculentum and Fagopyrum tataricum genotypes22,24,74. The present study could only give an insight into the genetic diversity and population structure from the analyzed set. Further research on genomic selection and desirable gene mining for specific traits for genetic improvement and yield augmentation of buckwheat need to be investigated in future. The lack of buckwheat genomic resources is also one of the reasons of slow progress in molecular breeding of buckwheat75.

Biochemical characterization of buckwheat genotypes

A core set of ten extreme genotypes was taken with respect to the yield parameter for further biochemical characterization for traits like total protein, total carbohydrate, and phytic acid. ANOVA for these three biochemical traits revealed the presence of significant and considerable variation among the genotypes under study. The selected genotypes showed significant variation in total protein, carbohydrate, and phytic acid content. In the tataricum species, all three nutritional traits were found to be greater than in the esculentum species. All three nutritional parameters had a high correlation with each other. Current findings were in line with4, who found a total protein concentration ranging from 5.65 to 14.17%. The protein content of three species of buckwheat germplasm, as observed by76, varied widely, with a range of 9.22–14.55%, higher than current results. It was also revealed that tartary buckwheat contains between 60 and 68% carbohydrates77. Likewise, it was reported that the range of phytic acid in different buckwheat products as 0.9–2 g/100 g78, which was on par with the findings of the current study. Hence, taking into account the nutritional content with yield and stability, the following accessions namely IC16550, IC26586, Shimla B-1, and Local teethey were identified as most promising from a set of 102 buckwheat accessions.

Conclusion

Genetic diversity and systemic characterization of that diversity is a key to effective genetic resources management and successful breeding. The study revealed deep insights into quantitative, qualitative and molecular variations that exist among the 102 buckwheat genotypes. The nature of population differentiation revealed the prevalence of among individuals’ variation within the population indicating the diverse nature of genotypes. Based on AMMI yield stability analysis, accessions like IC16550, IC42424, IC26606, EC18049, and Shimla B-1 were identified as most promising. While, based on multi-trait stability index values, fifteen genotypes, namely, IC37275, IC26592, Local Teethey, Shimla B-1, IC13139, IC36919, IC26606, IC26586, IC107987, IC107974, IC13412, IC13145, IC37269, IC107608, and IC26595, were chosen with a selection intensity of 20%. Out of the 102 accessions, four accessions namely IC16550, IC26586, Shimla B-1, and Local Teethey were identified as most promising in terms of yield, stability and nutritional content. These selected genotypes would be of immense value for genetic improvement breeding of buckwheat, yield augmentation and quality enhancement with special reference to northeastern hill region of India. The present investigation couldn’t elaborate in the genes/QTLs level of traits yield and key trait of the accessions. Hence, further study on exploitation of such promising accessions for QTLs/genomic level for trait specific need to be carried for varietal development program.