Circularity showed only small to medium significant correlation with seed width, minimum, and mean and median Gray Value . Integrated density had a negative correlation with the rest of the variables, statistically significant only to height and width. Partitioning of the variance and grouping of the accessions, regarding seed morphology, were estimated by hierarchical clustering under the Unweighted Pair Group Method with Arithmetic mean and Principal Component Analyses . The UPGMA analysis gave formation to a three cladded dendrogram , with most of the landraces grouped together in one clade, along with some of the cultivars that were separated in subclades. A particular landrace, GR2, presenting unique seed morphology formed a single clade alone. A third clade consisted exclusively of cultivars and breeding lines. A PCA performed to calculate the Eigenvalues and variability partitioning to thedifferent vectors, with the first two dimensions explaining 61.92% of the data variability.
In a bi-plot graph of the PCA, the genotypes that were found to be highly correlated, are in close proximity with one another, and the distribution of the accessions forms two groups, with GR2 being an outlier . Regarding the seed surface, landraces GR24, GR25, GR27, and GR28 have biconcave seeds of orange-white color, with dark flecks and a characteristic black line peripheral to the hilum, while the rest of the accessions have white-yellowish seeds with no spot pattern . Interestingly the partitioning of the 16 alkaloid compounds assessed in each sample varied between accessions, indicating that each alkaloid biosynthesis product may accumulate in different amounts in different genotypes . Furthermore, landraces GR24, GR25, GR27, and GR28 with spotted seeds had no higher alkaloid content compared to those with white-yellowish seeds. Further studies implementing more specific markers of alkaloid biosynthesis pathways are required to complement systematically the seed alkaloid profile and the genetic alleles configuration of the genotype. Moreover, the partitioning of the 16 alkaloids in the metabolic profile of landraces is quite diverse indicating that a more thorough assessment of critical alleles that impart accumulation of alkaloids is necessary for an integrated assessment of valuable germplasm that could be underscored in the sole absence of the pauper allele.
Genetic diversity of white lupin genetic resources in Greece was assessed using morphological and molecular markers. This is the first report for molecular characterization of local white lupin germplasm in a country that is considered as a center of genetic variation for this crop. Seed morphological characteristics were selected because heritability of seed morphological traits and ease of access to seeds may provide a reasonable alternative to estimate the distribution of genetic variation . Considerable variation was recorded for TSW, seed size, and coloration while less variation could be detected for circularity. Clustering of genotypes formed subclades consisted solely of landraces, although some landraces were partitioned with the cultivars. Other studies indicate high level of polymorphism for selected morphological traits in white lupin, while generally these characters are not very informative for landrace genetic diversity studies. Wide range of variation was recorded for proportion of pod walls and other pod components in 325 ecotypes originated from 17 countries of the Mediterranean region and north Africa. The variation was related with the country of origin of the accessions since Egyptian ecotypes had the lowest proportions of pod walls while the Greek and Italian ecotypes had the highest . Analysis of Ethiopian landraces indicated the existence of high level of polymorphism for different agronomic traits and nutrient contents of grain . In an extensive study of 35 Spanish white lupin accessions estimating variation for 50 quantitative and 51 qualitative characters, described in the literature as highly heritable, only a small number of three quantitative and one qualitative parameters were variable enough to provide good separation of the accessions .
It is conceivable that as valuable as they may be, agronomic traits and morphological characteristics as genetic markers are of limited resolution power for genetic diversity assessment, compared with molecular markers. In this study, molecular polymorphism was considerable. Relatively moderate to high PIC values, indicating a high level of genetic diversity in the germplasm assessed. The markers were suitable for discrimination of most landraces that clustered apart from the commercial cultivars indicating a unique diversity. Specifically, landraces were grouped into two clusters with those originated from Andros island forming a distinct group representing the wild graecus form of lupin, while all the others were grouped together regardless of geographic origin. Interestingly, all the commercial cultivars were grouped in one cluster. Molecular markers have been proved valuable tools for germplasm diversity studies in many plant species. Yet, limited number of studies on white lupin molecular diversity are cited in the literature. To the best of our knowledge, genetic diversity using molecular markers has been estimated for local germplasm from Ethiopia and Egypt .