Introduction

The Medicago genus consists of approximately 90 species worldwide, predominantly found in Western Europe, Western Asia, Southeast Asia, and Africa. Most species within this genus are wild herbs, with many being significant as forage crops1. In China, Medicago species are primarily distributed across 14 provinces, especially in regions north of the Yangtze River and northwest of the Yellow River. Notable species, such as Medicago sativa L. (alfalfa), Medicago falcata L., and Medicago polymorpha L., are recognized for their high yield, palatability, and nutritional value2. Wild Medicago varieties in China, such as M. edgeworthii L., are notable for their strong cold tolerance. For example, M. edgeworthii thrives in high-altitude, cold regions like Tibet, Qinghai, and Yunnan, often growing along hillsides3,4. Similarly, M. archiducis-nicolai L. is found on slopes and grasslands at elevations of 2000–4000 m in areas like Qinghai, Gansu, Ningxia, and Tibet. In contrast, M. polymorpha L. is less cold-tolerant and is primarily found in the warmer, more humid regions of Jiangsu, Zhejiang, Sichuan, Hubei, Hunan, and Guizhou5,6. As leguminous forage plants, Medicago species are not only valuable for their nitrogen-fixing abilities but also exhibit tolerance to cold temperatures, making them integral to agricultural ecosystems. They contribute significantly to nitrogen cycling, soil and water conservation, soil fertility enhancement, and soil structure improvement7. Among these species, M. sativa (alfalfa) is the most important forage crop8,9. Often referred to as the “king of forage,” alfalfa is a perennial herbaceous plant rich in minerals such as calcium, phosphorus, potassium, magnesium, and sulfur, as well as essential vitamins10,11. Cultivated for over 2000 years in China, alfalfa is now grown globally12. In the United States, it ranks as the fourth most valuable crop after Zea mays L., Glycine max L., and Triticum aestivum L13. Its popularity stems from its high yield, rich protein content, good palatability, and adaptability. In recent years, with rapid economic development and rising living standards, demand for livestock products in China has surged, leading to a steady expansion of alfalfa cultivation areas14.

In our experiment, the Trigonella genus, a related group to Medicago, was included for comparison. The Trigonella genus contains approximately 70 species globally, fewer than those in Medicago15,16. Most Trigonella species are found in regions along the Mediterranean coast, central Europe, northern Africa, southwest and central Asia, and Oceania17,18. In China, only nine species of Trigonella have been identified, with Trigonella foenum-graecum L. (fenugreek) being the most prominent. Fenugreek is an annual or biennial herb known for its diverse uses19. It is widely recognized for its medicinal properties, such as dispelling cold, alleviating dampness, and relieving pain20. Additionally, fenugreek seeds contain diosgenin, a steroidal compound that plays a critical role in the synthesis of hormone-based medicines, demonstrating significant pharmaceutical potential21,22,23. Furthermore, studies have shown that fenugreek contains 4-hydroxyisoleucine, which has demonstrated efficacy in managing Type II diabetes24. Fenugreek gum is also used in the oil refining process for preparing fracturing fluid and as a thickening agent in the food industry25. Moreover, processed fenugreek is used in perfumes, further expanding its commercial applications. As demand for fenugreek increases, research into its chemical composition, pharmacological properties, cultivation methods, and diverse uses continues to grow.

As a closely related genus to Medicago, Trigonella shares several similarities with Medicago species. Both are leguminous forage plants that are nutrient-rich and have valuable wild and cultivated varieties for forage use. Some species from both genera thrive in similar environments and exhibit comparable morphological characteristics, making them difficult to differentiate. Numerous studies have demonstrated a close genetic relationship between Medicago and Trigonella, with transitional groups blurring the lines between the two genera. According to available agronomy literature, the seed size difference between plants of the Trigonella genus and those of the Medicago (Alfalfa) genus varies depending on the species and environmental conditions, however, in general, Medicago species tend to have larger seeds than Trigonella species. Seeds of Trigonella plants are typically smaller in both diameter and mass. For instance, the seeds of Trigonella are approximately 1.5 millimeters in diameter, whereas seeds of Medicago sativa (alfalfa) typically range from 2 to 3 millimeters in diameter26,27,28. These intermediate species, often referred to as “Medicago-like” or “Trigonella-like,” include species such as Melilotoides ruthenica L29. Despite these similarities, the taxonomic classification of these intermediate groups remains unclear, and no comprehensive study comparing the two genera has been conducted to date.

The study of morphological seed diversity focuses on both quantitative and qualitative traits, collectively referred to as phenotypic traits30,31. These traits reflect the morphological diversity of seeds and are of great value in biological research, plant breeding, and other fields. As research into morphological diversity has gained momentum, it has become an increasingly prominent topic both in China and globally32. Various experiments have been conducted to measure and analyze the phenotypic characteristics of germplasm accessions. For example, Alireza and Ali (2020) analyzed the phenotypic traits of Damask rose germplasm accessions, contributing to the selection of superior breeding lines33. Dong (2016) assessed phenotypic traits in 537 common vetch germplasm accessions, revealing significant diversity that aided taxonomic studies34. Similarly, Oral (2020) observed high morphological variability in Caribbean breadfruit germplasm, with quantitative and qualitative traits providing insights into these variations. This research identified superior germplasm accessions with increased morphological variability and phenotypic diversity, supporting the commercial production and utilization of breadfruit35. Thus, studying the morphological diversity of germplasm resources is of great importance. However, few studies have simultaneously analyzed such a wide range of species.

In this experiment, 587 Medicago and 32 Trigonella germplasm accessions were collected from germplasm banks in China and the United States. The large sample size, with multiple accessions per species, provided a robust dataset. High-resolution images of seeds were captured, and 12 seed characteristics-straight length (SL), straight width (SW), width-to-length ratio (WL), perimeter (PE), radicle length (RL), hilum length (HL), 100-seed weight (SY), seed size, seed coat condition, seed shape, seed color, and radicle characteristics-were statistically analyzed. The findings provide valuable data for future efforts to identify and evaluate seed morphology in Medicago and related genera.

Results

Analysis of quantitative and qualitative characters

A total of seven quantitative and five qualitative traits from 587 Medicago and 32 Trigonella germplasm resources were analyzed. Their accession numbers can be checked in Table S1. The results of the quantitative analysis are presented in Tables S2 and S3, while representative seeds for different quantitative traits are shown in Fig. 1.

Fig. 1
figure 1

Representative photos of germplasm accessions of different quantitative traits, including seed coat, seed color and seed shape. The white measuring scale represents 2000 μm.

Full size image

For easier figure preparation, we have assigned numbers to the seeds of 77 Medicago species and 21 Trigonella species (Table 1). The results in Table 2 indicate that, among the 77 Medicago species, the average seed length (SL) was 2.64 mm, average seed width (SW) was 1.64 mm, average seed perimeter (PE) was 7.00 mm, average hilum length (HL) was 0.09 mm, average radicle length (RL) was 0.96 mm, and average 100-seed weight (SY) was 0.31 g. The seed perimeter exhibited the largest standard deviation (1.82 mm), while the largest coefficient of variation (CV) was observed for 100-seed weight (98.69%), followed by hilum length. The smallest CV was found in the width-to-length ratio, at 20.90%.

Table 1 Germplasm code of Medicago and Trigonella Genus.
Full size table
Table 2 Statistical analysis of the quantitative characters among Medicago species.
Full size table

Further analysis, as shown in Table 2, reveals variation within each species of the Medicago genus. For instance, the average SL of Medicago sativa (accession no. 62) was 2.14 mm, with a CV of 14.19%. The largest CV in M. sativa was observed for HL (32.28%). Comparing these results with those in Table S2, it is evident that there were generally larger CV values observed between species.

In Table S3, the results show that, among 21 different species in the Trigonella genus, average seed length was 2.31 mm, average seed width was 1.47 mm, average seed perimeter was 6.12 mm, average hilum length was 0.08 mm, average radicle length was 1.11 mm, and average 100-seed weight was 0.33 g. The largest value of standard deviation was seed perimeter, which was 1.94. The highest coefficient of variation was 100-seed weight, which was 112.98%, followed by hilum length and perimeter.

A coefficient of variation (CV) greater than 10% indicates a substantial difference between the samples. The CV for seed length, width, width-to-length ratio, perimeter, hilum length, radicle length, and 100-seed weight in Medicago germplasms ranged from 20.90 to 98.69%, with an average CV of 40.00%. For Trigonella, the variation coefficients for these traits ranged from 23.98 to 137.45%, with an average CV of 48.77%. The largest CV was observed for 100-seed weight in both genera. These results suggest that the CV of Trigonella was larger than that of Medicago, and that both genera exhibited substantial variation in their quantitative traits. Additionally, variance analysis revealed significant differences (P < 0.05) among traits, confirming that there were considerable variations in quantitative characteristics between the 587 Medicago and 32 Trigonella germplasms, reflecting rich morphological diversity.

The statistical results for the qualitative traits of Medicago germplasms are shown in Table 3. Seed coats were classified into four types: smooth, folded, and those with black or white spots. Most germplasms had smooth seed coats (83.48%), followed by folded seed coats (14.31%), with black and white spot patterns being rare. Seed shape was categorized into six types: reniform, obround, circular, irregular, rhombic, and rod-shaped. Reniform seeds were the most common, comprising 62.01% of the total. Circular, rod-shaped, and rhombic seed types were less frequent. Seed color was divided into five categories: brown, black, brownish yellow, gray, and yellowish green. Brownish yellow seeds were the most frequent (75.98%), followed by brown seeds (14.48%), with other colors occurring in smaller proportions. Radicle length was classified into two categories: long and medium, with medium radicles representing 72.91%, and long radicles accounting for 18.74%. Medium-sized seeds made up 84.33% of the total.

Table 3 Statistical analysis of the qualitative characters of Medicago germplasm resources.
Full size table

The statistical results for the qualitative traits of Trigonella are presented in Table S4. Seed coats were categorized into three types: smooth, folded, and those covered with black spots. The highest proportion of germplasms had smooth seed coats (50.00%), followed by those with black spots (34.38%), while folded seed coats were the least common (15.63%). Seed shape was classified into five types: obround, circular, rod-shaped, rhombic, and irregular. Obround seeds were most prevalent (34.38%), with circular and irregular seeds each comprising 25.00%. Seed color was classified into six categories: brownish yellow, purple, brown, gray, green, and yellowish green. Brownish yellow seeds were most common (56.25%), followed by yellowish green and brown seeds (15.63%). Medium-length radicles were the most frequent (46.88%), and medium-sized seeds represented 43.75% of the total.

Cluster analysis

Cluster analysis of quantitative characters of Medicago and Trigonella was carried out by using R language software. The results are shown in Fig. 2 and S1.

Fig. 2
figure 2

Dendrogram of Medicago germplasm. The 77 different species of Medicago genus can be roughly divided into four groups according to their SL, SW, PE, and SY, signified by four parts surrounded by red lines.

Full size image

The 77 different Medicago species can be divided into four groups according to their quantitative traits. There are four different germplasms in Medicago surrounded by red frames in Fig. 2, and seeds are larger from the left to the right side.

The 23 different species of Trigonella species are roughly divided into two groups according to their quantitative traits that were measured before (Fig. S1). The result may not be accurate, cause the number of Trigonella germplasms was far less than that of Medicago germplasm.

Principal-component analysis

Principal component analysis (PCA) was conducted using the seed traits of Medicago and Trigonella germplasms, including seed length (SL), seed width (SW), width-to-length ratio (WL), radicle length (RL), hilum length (HL), and 100-seed weight (SY), to clearly illustrate the contribution of each trait to morphological seed diversity. The correlation analysis (Table S5) revealed that most quantitative traits were positively correlated, except for WL, SL, perimeter (PE), and SY, which were negatively correlated. The strongest correlations were observed between SL and PE (0.990), and SW and PE (0.944).

In the PCA, components with eigenvalues greater than 1 were selected to ensure that each component provided meaningful interpretation and representativeness. Three principal components were extracted, explaining a cumulative variance of 85.64%. The loadings of each trait under the different principal components reflect the degree to which each trait contributes to the component. These results are summarized in Table S6.

From the PCA results, in the first principal component, SL, SW, HL, RL, and SY had the largest loadings, collectively representing the primary quantitative traits of Medicago germplasms. In the second principal component, the width-to-length ratio (WL) had the highest loading, supplementing the quantitative characteristics of seed morphology. The third principal component was dominated by seed coat, shape, color, and radicle traits, which are key qualitative traits of the germplasm.

The biplot (Fig. 3) shows 7 quantitative characters in 8 representative germplasms selected in Fig. 2, which were divided into four groups according to their SL, SW, PE, and SY. We randomly selected two different germplasms in each group to make PCA-plot, and they can be divided into four groups, too.

Fig. 3
figure 3

PCA-plot of 7 quantitative characters in 8 representative germplasms selected in Fig. 2, which were divided into four groups according to their SL, SW, PE, and SY. We randomly selected two different germplasms in each group to make PCA-plot, and they can be divided into four groups, too. The result was the same as the diagram. Group 1 contained M. astroites and M. fischeriana; Group 2 contained M.sativa and M. rigidula; Group 3 contained M. radiata and M. cretacea; Group 4 contained M. intertexta and M. ciliaris.

Full size image

Multiple regression analysis and path analysis

Although we observed that seed length (SL), seed width (SW), hilum length (HL), radicle length (RL), and 100-seed weight (SY) had large loadings in the first principal component of the PCA, we sought to better understand how the other quantitative traits influence SY. The correlation results in Table S4 were not sufficient for this purpose. Therefore, we conducted path analysis, setting SY as the dependent variable (Y) and SL, SW, width-to-length ratio (WL), perimeter (PE), HL, and RL as independent variables, denoted as X1, X2, X3, X4, X5, and X6, respectively.

The path analysis results, presented in Table S7, show the following regression equation:

Y=-0.517 + 0.219 × X1 + 0.235 × X2–0.138× X3–0.022 × X4 + 1.351 × X5 + 0.31 × X6.

From this equation, we observe that SY was positively correlated with SL, SW, HL, and RL. Specifically, for every 1 mm increase in SL, SY increased by 0.219 mm; for every 1 mm increase in SW, SY increased by 0.235 mm; for every 1 mm increase in HL, SY increased by 1.351 mm; and for every 1 mm increase in RL, SY increased by 0.31 mm. In contrast, PE had a negative influence on SY, decreasing by 0.022 mm for each unit increase in PE.

The results of the regression analysis, shown in Table S8, indicate that the F-statistic was 68.366 with a P-value of less than 0.05, suggesting that the regression model is statistically significant (P < 0.05).

Boxplot of quantitative traits in Medicago and Trigonella species

A boxplot was constructed based on the numerical values of the quantitative traits of Medicago germplasms, with the results shown in Fig. 4.

Fig. 4
figure 4

Boxplot of 6 quantitative traits in Mediacgo and Trigonella genus. (A) Seed length, (B) seed width, (C) width to length ratio, (D) perimeter, (E) hilum kength, (F) radical length.

Full size image

From the statistical data presented in Fig. 4, it is evident that seed samples from the Medicago genus are generally larger than those from the Trigonella genus, particularly in terms of seed length, width, perimeter, and hilum length. However, when examining the width-to-length (WL) ratio, Trigonella seeds exhibit a larger proportion compared to Medicago seeds. This suggests that, while Medicago seeds may be larger in size overall, Trigonella seeds stand out for their distinctive seed shape in terms of the WL ratio. Additionally, there is little difference in radicle length between the two genera, making this trait less useful for distinguishing them. Furthermore, the variation or dispersion in the six quantitative traits measured for Trigonella seeds is more pronounced compared to Medicago. This greater variability may be partly due to the smaller sample size for Trigonella seeds, which could influence the reliability of the calculated averages and standard deviations for these traits.

According to statistical data (Fig. S2), the average straight length of the 77 Medicago species was 2.64 mm, indicated by the dotted line in the boxplot. In addition, the boxplot showed that the length of M. scutellata (no.65), M. doliata (no.19), and M. intertexta (no.29) were significantly different from the others. The seed length of M. falcata (no.21) was smaller than that of M. sativa (no.62).

According to the statistical data (Fig. S2A), the average seed width of Medicago germplasm was 1.64 mm. The boxplot showed that the widths of M. scutellata and M. doliata were much larger than the average, while the widths of M. astroites (no.4) and M. fischeriana (no.22) were smaller than the average.

According to the statistical data (Fig. S2B), the average perimeter of Medicago was 7.00 mm. The box plot showed that the perimeter of M. scutellata was much larger than the average. Besides, M. turbinate (no.75) had huge difference within species, because it had the longest bar in boxplot.

According to the statistical data (Fig. S2C), the average length of the whole hilum was 0.090 mm. It can be seen from the box plot that the hilum lengths of M. scutellata and M. intertexta were much larger than the average, while the hilum length of M. biflora (no.5) was significantly smaller than the overall average.

The average radicle length of Medicago germplasms was 0.960 mm. According to the boxplot (Fig. S2), the radicle length of M. intertexta (no.29) was much larger than the average.

Considering the length and completeness of this article, we also prepared several boxplots of different quantitative traits in the Trigonella genus, these pictures can be found in the supplementary materials (Fig. S3).

Discussion

The diversity observed in both qualitative and quantitative traits provide valuable insight into the morphological characteristics of plants. This morphological diversity reflects the genetic and phenotypic richness inherent in germplasm collections, which are essential resources for plant breeding efforts. As noted by Wonneberger (2017)36, the variety within germplasm collections forms the foundation for successful breeding programs, offering a broad range of traits that can be selectively combined to achieve desirable crop characteristics. In this study, we analyzed 587 Medicago and 32 Trigonella germplasms sourced from Chinese and American germplasm banks. The selected samples were subjected to an extensive statistical analysis of 12 key morphological traits, including seed length (SL), seed width (SW), width-to-length ratio (WL), seed perimeter (PE), radicle length (RL), hypocotyl length (HL), seed yield (SY), seed coat characteristics, seed shape, seed color, seed size, and radicle condition. Our findings revealed substantial morphological diversity within both genera, highlighting the wide range of phenotypic expressions present in these germplasm collections. This rich diversity underscores the potential of these germplasms to significantly contribute to breeding programs. By investigating and harnessing these unique traits, researchers and breeders can select for important attributes-such as disease resistance, drought tolerance, and improved yield-ultimately enhancing the genetic pool available for agricultural advancement.

In our experiment, the variation range of seed length (SL) was 1.01 to 5.84 mm in the Medicago genus and 0.93 to 5.92 mm in the Trigonella genus. For seed width (SW), the variation range was 0.51 to 3.89 mm in Medicago species and 0.63 to 3.89 mm in Trigonella species. The variation range for seed yield (SY) in the two genera was 0.06 to 1.84 g for Medicago and 0.04 to 2.25 g for Trigonella. Notably, the coefficient of variation for 100-seed weight was the largest across all traits. Additionally, there were differences in the coefficient of variation for 100-seed weight within each species. For Trigonella, the coefficient of variation exceeded 100%, likely due to the smaller sample size of Trigonella germplasms, which may have resulted in a larger deviation.

Seed size plays a critical role in seed reproduction and seedling establishment. Research suggests that, given equal external energy, plants with smaller seeds tend to produce more seeds and exhibit stronger reproductive capacity. In contrast, larger seeds often show better stress resistance, particularly during the later stages of seedling establishment, when environmental stress factors are more pronounced37. This highlights the importance of seed morphology in influencing seedling development, survival, and establishment in varying environments. Thus, information on seed size and related traits in plant germplasm can be instrumental in variety selection, particularly for improving seed yield, seedling establishment, and survival rates38. In this experiment, the Medicago and Trigonella seeds were collected from 19 different countries, each exhibiting distinct qualitative and quantitative traits. By combining environmental factors with these morphological characteristics, we can offer more accurate data for seed identification and classification, facilitating more precise breeding and selection strategies based on the unique traits of each germplasm source.

The morphological diversity observed in plants is the result of the complex interaction between genes and the external environment. Phenotypic traits, which are influenced by gene expression and regulation, often provide valuable insights into the underlying genetic diversity. These phenotypic variations are not only a product of long-term evolutionary processes but also serve as a foundation for the continued adaptation and evolution of germplasms in specific environmental contexts27. Thus, studying morphological diversity is crucial for understanding the evolutionary trajectories of species or populations and offers important insights into their breeding potential39. In our study, we examined germplasms from the Medicago and Trigonella genera. Analysis of the data revealed that while these germplasms exhibit broad geographic distribution, notable differences exist between species within the same genus. These differences reflect varying levels of adaptability to environmental conditions, emphasizing the role of both genetic and environmental factors in shaping phenotypic diversity. Such variation underscores the importance of considering both genetic and environmental influences when studying species diversity and breeding potential.

In this experiment, we observed substantial morphological diversity among the different germplasms of Medicago and Trigonella, suggesting that each germplasm carries unique genotypes shaped by long-term evolution and adaptation to diverse environments. This diversity has significant implications for studying species evolution. The phenotypic variation observed within a genus is primarily due to intrageneric variation, which results from the complex interaction between genetic factors and environmental conditions. Phenotypic diversity often mirrors genotypic variation to a certain extent40. For example, Medicago species are distributed across the Mediterranean, southwestern and central Asia, and Africa. Over time, Medicago has evolved new genotypes to adapt to these varied environments, leading to the formation of distinct phenotypic traits, which have formed the basis for breeding programs.

While molecular markers, such as simple sequence repeats (SSR), have shown significant genetic similarity between Trigonella and Medicago41, this experiment focused primarily on phenotypic traits. The cluster analysis revealed similarities between the two genera in key quantitative traits such as seed length, width, and perimeter. This suggests that distinguishing between Medicago and Trigonella based on these morphological traits alone can be challenging, highlighting the need for more comprehensive analysis, including molecular techniques, to better differentiate them.

Principal component analysis (PCA) has been applied to various crops, such as oats42 and soybeans43, to simplify the analysis and comprehensively capture the morphological characteristics of plant populations. In this experiment, PCA was performed on 587 Medicago germplasms, extracting three principal components that explained a cumulative 85.64% of the total variance. In the first principal component, traits like seed length (SL), seed width (SW), hilum length (HL), root length (RL), and seed yield (SY) had higher loadings, reflecting the overall quantitative characteristics of Medicago germplasms. The second principal component highlighted the seed width-to-length ratio, which served as a supplementary trait to the quantitative seed features and complemented the qualitative characteristics of the germplasm. In the third principal component, traits such as seed coat, seed shape, seed color, and radicle condition were more influential, providing additional insights into the qualitative aspects of the germplasm. These findings suggest that seed size can be an effective index for simplifying the seed collection process, making it more streamlined and accurate. By focusing on key traits revealed through PCA, researchers can optimize germplasm evaluation and improve the efficiency of field assessments.

Germplasm resources are crucial for advancing both taxonomic and breeding research, as they provide the genetic diversity essential for understanding species classification and for developing improved crop varieties44,45,46. However, current taxonomic research often encounters challenges due to incomplete species representation, limited genus coverage, and the underrepresentation of key species in certain contentious groups47,48. This gap in comprehensive germplasm data hampers the resolution of many taxonomic issues. For instance, in the Medicago genus (commonly known as alfalfa), several groups contain only a single species, despite their critical taxonomic importance49. These groups are underexplored in research, which increases the risk of overlooking valuable taxonomic insights. Therefore, a global collaborative effort among alfalfa taxonomists is essential to comprehensively collect and share germplasm resources50. Such efforts would strengthen the foundation for addressing these taxonomic gaps51. One example is the Sporocarps group within the Medicago genus, which comprises 35 species and represents one of the most complex and poorly understood evolutionary relationships in the genus52. The genetic diversity within this group is highly variable, and a more focused collection and preservation effort is needed to clarify its evolutionary history53,54. Moreover, the presence of species like M. medicaginoides and M. orbicularis in the wild remains uncertain and warrants further field investigations55. From a breeding perspective, there are three primary objectives that should guide research on alfalfa germplasm resources. First, there is an urgent need for the extensive collection of wild germplasm, particularly focusing on species closely related to cultivated alfalfa, including various subspecies, alfalfa complex variants, and other Medicago species56,57. Utilizing high-throughput sequencing technologies will enable researchers to identify and locate valuable genes in both cultivated and wild species58. These genetic resources can then be incorporated into molecular breeding programs to enhance the resilience and productivity of high-yielding alfalfa varieties, especially for herbivorous livestock farming59,60. Secondly, while 19 species of Medicago are known to have cultivated varieties, further domestication efforts are required to improve their yield and quality. Such efforts offer significant promise for breeding programs aimed at increasing the economic and nutritional value of alfalfa in animal husbandry61,62. Third, wild alfalfa varieties, especially those native to specific regions, exhibit strong adaptability to local climates and soils. By collecting and domesticating these local germplasm resources, breeders can select for traits such as higher seed yield and lower hard seed rates. This would not only improve local livestock production but also support ecological conservation initiatives63. Areas like the Qinghai-Tibet Plateau, Xinjiang, and Inner Mongolia, which are rich in wild alfalfa resources, offer great potential for such targeted breeding efforts. In these regions, breeding could enhance agricultural productivity and contribute to sustainable land management practices64,65. Collectively, these taxonomic and breeding initiatives underscore the critical importance of germplasm resources in addressing research gaps, improving crop resilience, and contributing to sustainable agricultural and environmental practices66.

In this experiment, a substantial collection of Medicago and Trigonella germplasm accessions was analyzed. However, variations in the number of samples collected across different species introduced some inconsistencies that may have impacted the results. These discrepancies highlight the need for a more balanced representation of germplasms from each species to ensure robust and reliable findings. Looking forward, we aim to expand the germplasm collection to improve the comprehensiveness and accuracy of our dataset, which will enable more nuanced and representative analyses across a wider range of species. Additionally, the current analysis of morphological diversity was limited to 12 phenotypic traits. While these traits provided valuable insights, there remains significant potential to expand our understanding by investigating additional seed characteristics. Exploring a broader spectrum of morphological and genetic traits would allow for a more complete and detailed evaluation of Medicago and Trigonella seeds, thereby enriching the information available for their identification, characterization, and assessment in breeding programs. Future studies will thus aim to incorporate a wider array of phenotypic markers, enhancing our ability to identify and evaluate unique traits within these genera and supporting their use in agricultural innovation.

Materials and methods

Seed materials

In this study, a total of 619 germplasm accessions were evaluated. These germplasms consisted of 587 accessions in the Medicago genus, which contains 77 different species, and 32 accessions in the Trigonella genus, which contains 21 different species. These accessions were acquired from the National Animal Husbandry Station of China and the National Plant Germplasm System of the United States. They were from 19 countries, including China, the United States, Russia, German, Australia, and Canada. These seeds were collected in different years. For convenience, a code table was prepared (Table 1) to represent every germplasm in the Medicago and Trigonella genera. We selected representative seed photos which were obtained by the Discovery.V20 instrument produced by Zeiss, Germany (Fig. 5). We uploaded seed photos for each species, that can be found on the web: https://pan.baidu.com/s/19n8chQzNy6x2awv-n5YsNA, the password was “Medi”, in a file titled “Seed Photos”.

Fig. 5
figure 5

Representative photos of germplasm accessions of Medicago and Trigonella genus. Each picture represents different germplasm: (A) M. agropyreterum; (B) M. arborea; (C) M. borealis; (D) M. coerulea; (E) M. difalcata; (F) M. doliata; (G) T. suavissima; (H) T. stellata; (I) T. spruneriana; (J) T. spicata; (K) T. schlumbergeri; (L) Trigonella macrorrhyncha. There were large differences among the received germplasm resources. The white measuring scale represents 2000 μm.

Full size image

Morphological traits evaluation

In this study, 12 morphological traits, including 7 quantitative traits and 5 qualitative traits, were recorded. Qualitative traits were examined on the basis of seed coat, shape, color, size, and radicle situation. Seed coat was recorded for smooth, fold, and black and white spots. Seed color was recorded for brown, gray, black, green, purple, brownish yellow, and yellowish green. Seed size was recorded for large, medium, and small size. Seed shape was reniform, rhombic, circular, obround, rod-shaped, and irregular. The radicle situation was recorded for long, medium, and short radicles. If radicle length exceeded 2/3 of the width of the seed, the seed was classified as a long radicle. If radicle length was within 1/3 − 2/3 of seed width, the seed was classified as a medium radicle. The rest were classified as short radicles. To avoid the error of classification caused by different observers, the same observer was used in the statistics of a certain trait.

Quantitative traits were examined based on straight length (SL), straight width (SW), width-to- length ratio (WL), perimeter (PE), hilum length (HL), radicle length (RL), and 100-seed weight (SY). SL, SW, PE, HL, and RL were measured on Digimizer 4.2 software. WL was calculated on SPSS 21.0, and SY was measured by an electronic balance.

Statistical analysis

In this study, we used Discovery.V20 stereoscope produced by Zeiss, Germany to obtain seed images, which were produced in Digimizer 4.2 software. Then, we measured SL, SW, RL, and HL. The 100-seed weight was measured by an electronic balance with 0.01 g precision. Morphological evaluations were carried out using at least 20 repetitions of per accession.

The average values of the measured characters were used for analysis. The coefficients of variation (CV) were calculated as the variance index. Correlation between traits was determined using Pearson’s correlation coefficient with SPSS software version 21.0 (SPSS Inc., Chicago, IL, USA, Norusis 1998). The relationship between accessions was assessed using principal-component analysis (PCA) with SPSS 21.0 software. The first and second components (PC1 and PC2) were used to create the scatter plot with SPSS 21.0 software. Cluster analysis dendrogram was made in R 3.6.3. In addition, data of quantitative traits were used to produce boxplots; the tools that we used are on the following website: http://www.ehbio.com/ImageGP/index.php/Home/Index/Boxplot.html.

Conclusions.

The diversity of both qualitative and quantitative traits is essential for the morphological characterization of plants, and the genetic variability within germplasm forms the cornerstone of successful breeding programs. In this study, 587 Medicago and 32 Trigonella accessions were selected from germplasm banks in China and the United States. A statistical analysis of 12 traits-seed length (SL), seed width (SW), width-to-length ratio (WL), seed perimeter (PE), radicle length (RL), hilum length (HL), 100-seed weight (SY), and seed color-revealed significant morphological diversity within both genera.

While this study gathered a diverse range of germplasms, the number of accessions varied across species, which could affect the comprehensiveness of the findings. To strengthen the robustness of future research, increasing the number of accessions is recommended. Additionally, this analysis focused on 12 phenotypic traits, but a more in-depth exploration of additional seed characteristics would provide a more detailed and comprehensive understanding, aiding in the accurate identification and evaluation of Medicago and Trigonella seeds.

This research lays a valuable foundation for establishing seed banks and further expanding germplasm resources through a systematic examination of morphological traits.