Seed proteins variability
The three fractions electrophoregrams are represented by some accessions illustrated in Fig. 2. Figure 3 presents dendrograms generated using UPGMA and Jaccard’s index.
Albumins patterns
A total of 131 bands were detected with molecular weights ranged from 3.23 to 148.17 kDa. Each profile presents between 11 and 26 bands. All samples had more than one seed protein pattern. Intra-accessional diversity was also investigated by examining populations of the same taxa. The largest number of bands (25) is obtained in V. monantha subsp. calcarata, V. tenuifolia and V. sativa subsp. obovata. The lowest number (11) is obtained in V. lutea subsp. vestita. The band 37.73 kDa is the most common as it appears in 48 profiles followed by the band 9.19 kDa appeared in 43 profiles and the band 24.22 kDa observed in 40 profiles. In parallel, bands 3.23, 9.46, 15.81, 19.80, 24.87, 25.97, 46.86, 63.69, 66.34, 76.29, 99.86, 113.50 and 125.98 kDa are the least common as they are specific for one accession and appear each in 1 profile, followed by bands 8.85, 14.93, 16.72, 42.84, 75.43, 77.31, 82.63 and 88.47 kDa appeared each in two profiles. The cluster analysis indicated the discrimination into five groups at 0.86 Jaccard distance (Fig. 3). The first cluster can be divided into two groups, the first one includes accessions 12, 3, 27, 18, 26. The second one regroups samples 1, 58, 63, 87, 4, 90, 62 and 79. The cluster II is divided into 2 sub-clusters. The first one contains samples 5, 6, 59, 14, 52, 65, 36, 93, 7, 10, 64 and 86. The second one can further be divided into two groups: II2a contains accessions 34, 41, 55, 30 and 81. II2b is composed of samples 32, 66, 68, 57, 61. The cluster III regroups two sub-clusters. The first one contains accessions 33, 35, 11, 37, 83, 42, 38. The second one is further divided into two groups: III2a comprises sample 88 linked to 40, 29, 102, 49, 77, 45, 60, 74, 78, 44 and 91. III2b contains accessions 17, 28, 51, 20, 22, 23 and 84. The cluster IV comprises the sample 100 one side and accessions 72, 56, 70, 46, 43, 53, 71, 89, 19 and 80 another side. Finally, the cluster V contains the sample 98 (J = 0.88). The proximity matrix using Jaccard index shows that the higher distance (J = 1) is observed between the following couples: 17-4, 1-30, 1-32, 1-37, 1-41, 1-81, 4-28, 4-32, 4-84, 6-63, 6-87, 7-87, 63-10, 87-10, 14-63, 14-87, 17-63, 28-63, 30-58, 32-58, 62-32, 79-32, 32-90, 37-90, 58-37, 37-62, 41-58, 41-1, 52-87, 55-1, 55-58, 58-81, 66-58, 58-41, 58-30, 58-32, 59-63, 87-59, 63-93, 63-84, 63-7, 63-23, 63-52, 63-64, 63-65, 64-87, 65-87, 1-66, 79-32, 87-93, 87-65. The lower distance (J = 0.10) is obtained between 18 and 26. A distance of 0.16 is observed between 43 and 19. Between 89 and 19, a distance of J = 0.15 is observed. A distance of 0.17 is obtained between 90 and 62.
Globulin patterns
A total of 119 bands were obtained with molecular weights ranged from 2.77 to 131.88 kDa. Each profile presents between 6 and 27 bands. Two accessions (41 and 55) showed a unique protein pattern; the remaining accessions had more than one protein pattern. The largest number of bands (27) is observed for accessions 43 (V. monantha subsp. calcarta) and 87 (V. lutea subsp. eu-lutea). The lowest number (6) is obtained for samples 17 and 32 belonging to V. sativa subsp. obovata. The band 35.53 kDa is the most common as it appears in 52 profiles, followed by the band 49.45 kDa observed in 45 profiles and the band 33.44 kDa found in 37 profiles. In another side, bands 5.10, 5.46, 51.74, 67.97, 78.27, 85.47, 95.33, 100.77, 103.66, 104.69, 115.04, 119.96, 130.35 kDa are the less common as they appear each in one profile, followed by bands 16.86, 18.39, 19.11, 28.98, 37.69, 47, 38.87, 68.53, 86.70 and 131.88 kDa found in two profiles. Six major clusters were obtained at the distance of 0.87 (Fig. 3b). The cluster I is further divided into two sub-clusters (I1, I2). I1 includes the accession 46 linked to 62, 63, 90, 79, 88, 4, 87, 1 and 58. I2 can be divided into two groups. The first one (I2a) includes 59, 52, 65, 86, 64, 36, 14 and 93. The second one (I2b) contains samples 6, 7 and 10. The cluster II is divided into two subclusters (II1, II2). II1 comprises two groups. II1a contains accessions 55, 41, 30, 81, 34, 23. II1b is composed of 35, 15, 13, 8, 33, 85, 83, 42, 11, 37, 47, 38. II2 comprises sample 12 linked to 49, 60, 45, 74, 18, 29, 26, 40, 44, 3 and 27. Cluster III contains sample 43 linked to accessions 68, 80, 5, 72, 19 and 61. The cluster IV (J = 0.92) comprises samples 32, 17, 28, 22, 20, 51 and 57. The cluster V (J = 0.87) is divided into two sub-clusters. V1 includes 70, 56, 100, 71, 53 and 89. V2 contains samples 102, 78, 77, 84 and 91. The accession 98, being itself the cluster VI at J = 1. The dissimilarity matrix shows that a distance of 0.00 is observed between couples: 62-63, 55-41. A low distance of 0.11 is observed between samples 1 and 58. Between accessions 1 and 4, a distance of 0.14 is observed. Samples 58 and 4 are distant by J = 0.18. A distance of 0.14 is obtained between 1 and 87. Samples 4 and 88 have a distance of 0.15. The higher distance (J = 1) is observed between 98 and all other accessions. The same distance is observed for the following couples: 5-66, 5-34, 5-22, 5-23, 10-20, 10-22, 72-17, 68-17, 46-17, 43-17, 19-17, 81-19, 84-19, 66-19, 65-19, 55-19, 57-19, 51-19, 41-19, 32-19, 34-19, 23-19, 20-52, 22-80, 22-72, 22-68, 22-62, 22-63, 22-52, 22-46, 22-28, 22-19, 19-30, 32-43, 32-46, 32-68, 32-71, 46-51, 46-57, 51-68, 51-72, 57-72, 93-61, 100-65, 93-80, 98-100 and 98-102.
Prolamin patterns
A total of 98 bands were obtained with molecular weights ranged from 11.36 to 137.638 kDa. Each profile presents between 6 and 24 bands. Two accessions (1 and 4) showed a single protein pattern; the remaining accessions had more than one seed protein pattern. The largest number of bands (24) is observed in patterns 58 (V. lutea subsp. vestita) and 62 (V. lutea subsp. eu-lutea). The lowest number (6) is obtained for samples 19 (V. sativa subsp. angustifolia), 7 and 51 both belonging to V. sativa subsp. obovata. The band 17.65 kDa is the most common as it appears in 46 profiles followed by the band 41.49 kDa appeared in 39 profiles and the band 35.53 kDa observed in 29 profiles. Bands 12.44, 51.51, 53, 56.62, 78.57, 83.92, 89.03, 94.53, 104.51, 105.45, 115.08, 118.17 and 137.63 kDa are the less common as they appear in one profile each, followed by bands 29.49, 66.63, 68.33, 81.77, 85.10, 106.71, 109.03 and 127.97 kDa obtained each in two profiles. The UPGMA generated four major clusters at the distance of J = 0.89. The first cluster (I) is divided into two sub-clusters (I1, I2). I1 includes samples 81, 41, 55, 30, 34, 23 and 66. I2 can further be divided into two groups (I2a, I2b). I2a contains accessions 88, 62, 4, 1, 63, 87, 79, 90 and 58. I2b is composed of 74, 49, 27, 44, 60, 40, 45, 18, 29, 26, 3 and 12. The cluster II comprises two sub-clusters (II1, II2). II1 contains accessions 59, 7, 86, 65. II2 includes 36, 14, 93, 64, 52, 5 and 6. The cluster III (J = 0.92) comprises two sub-clusters (III1, III2). III1 is composed of accessions 71, 53 and 70 in the group III1a and accessions 85, 11, 38, 37, 83, 15, 35, 13, 33, 8, 47 and 42 in the group III1b. The sub-cluster III2 includes two groups. III2a contains accessions 80, 72, 68 and 19. III2b is composed of 22, 17, 61, 32, 57, 28, 20 and 51. The cluster IV includes samples 89, 56, 91, 77, 102, 78, 84, 43, 98, 46, 100 and 10. The proximity matrix using Jaccard’s index shows that a distance of J = 0 is observed between samples 1 and 4. The same distance is obtained between accessions 11 and 85. Between accessions 40 and 60, a distance of 0.05 can be observed. A distance of 0.08 is obtained between samples 26 and 29. A low distance of 0.10 can be observed for the couples: 1-63, 4-63, 23-34, 55-81, 60-74. Between samples 3 and 29, 63 and 87, a distance of J = 0.16 is observed. Samples 41 and 81, 63 and 79 are distant by 0.11. A distance of 0.17 is obtained for: 3-26, 26-18, 49-29. Samples 49 and 12 are distant by J = 0.19. A distance of 0.18 is observed between accessions 12 and 26 and accessions 26 and 49. Accessions 26-27, 26-74 and 45-49 are distant by J = 0.14. A distance of 0.15 is observed for accessions 27-49 and 30-81. Couples 29-60 and 29-45 have a distance of 0.13. The higher distance (J = 1) is obtained for a large number of couples as for: 1-84, 1-100, 3-65, 3-59, 3-32, 3-28, 3-20, 4-84, 4-100, 5-100, 5-80, 5-72, 5-68, 5-61, 5-57, 5-51, 5-43, 5-34, 5-32, 5-28, 5-23, 5-19, 5-20, 5-22, 5-17, 6-43, 6-102, 7-102, 7-100, 7-91, 7-84, 7-77, 7-78, 7-66, 7-61, 7-46, 7-51, 7-43, 7-32, 7-33, 7-28, 7-20, 7-22, 7-17, 7-13, 7-8, 8-19, 8-43, 8-46, 8-56, 8-59, 8-65, 8-77, 8-78, 8-84, 8-86, 8-91, 8-98, 8-100, 8-102, 10-61, 10-57, 10-32, 10-28, 10-20, 11-19, 11-43, 11-46, 11-56, 11-59, 11-65, 11-77, 11-78, 11-84, 11-86, 11-91, 11-98, 11-100, 11-102, 12-65, 12-61, 12-59, 12-51, 12-32, 12-28, 12-19, 12-20, 13-19, 13-46, 13-56, 13-59, 13-65, 13-77, 13-98, 13-100, 13-72, 13-68, 13-61, 13-46, 13-51, 13-43, 13-32, 13-28, 13-20, 13-22, 13-17, 15-59, 15-65, 15-86, 17-86, 17-64, 17-65, 17-59, 17-52, 17-36, 18-20, 18-28, 18-32, 18-51, 18-59, 18-61.
Mantel test
A Mantel test based on Pearson’s correlation was used to highlight correlations between the matrices of albumins (matrix A), globulins (matrix B) and prolamins (matrix C). The p value was calculated from the distribution of r(AB) using 10,000 permutations with the value of r(AB.C) = 0.3099. This test showed significant correlation between the three fractions studied since the calculated p-value (<0.0001) is below the significance level of alpha (0.05 = 5%). Concerning the correlation between ecogeography and seed proteins, r values were −0.0012, −0.0039 and 0.0166 respectively for albumins, globulins and prolamins. p-values are 0.8233, 0.9319 and 0.3689 respectively for the three fractions. Thus, Mantel test showed no significant correlation between ecogeography and protein patterns since the calculated p-values are below the significance level of alpha.
Cluster analysis based on ecogeographic data
The dendrogram illustrated in Fig. 4 shows the relationships between these taxa, based on the variation in the five ecogeographic parameters studied. At the Euclidean distance of 716.43, the dendrogram can be divided into two major clusters (I and II). The first one is further divided into two sub-clusters (Ia and Ib). Ia (d = 119.65) comprises samples 44, 66, 41, 40, 26 and 70 belonging to 2 bioclimates (LH–SH). Ib can be divided into Ib1 (d = 226.96) and Ib2 (295.93). Ib1 contains the sample 77 (HA) linked to accessions 83, 89, 13, 8, 34, 23, 55, 58, 51, 10, 43, 78, 100, 84, 81, 102, 4, 47, 19, 32, 17 and 6 from 4 bioclimates (HA–MSA–SH–HSA). Ib2 (d = 295.93) is composed of samples 91, 98, 74 belonging to 2 bioclimates (SH–HSA) linked to 46, 18, 45, 27, 29, 3 (MSA) then 22, 20, 60, 80, 72, 57, 49, 42 and 86. (MSA–LSA–HSA–SH). The second cluster comprises subclusters IIa which contains samples 12 (LSA) and 11 (HA) and IIb. The latter comprises groups IIb1 (d = 165.56) composed of accessions 71, 38, 88, 87, 93, 28 and 59 belonging to 1 bioclimate (LH) and IIb2 which can further be divided into two groups: At a distance of d = 172.22, the first group contains accessions 64, 1, 63, 85, 33, 36, 62, 79, 61, 90, 15, 35 and 14 collected from two bioclimates (LH–SH). The second group (d = 317.53) comprises sample 65 (LH) linked to accessions 56, 68, 52, 37, 30, 7 and 5 (SH–MSA–SA). The higher distance (d = 1286) is observed between 91 (V. monantha subsp. cinerea, SH) and 93 (V. sativa subsp. consobrina, LH). A distance of d = 1284 is obtained between 87 (V. lutea subsp. eu-lutea, LH) and 91. The distance of d = 0 is obtained between the following couples: 35-14, 6-17, 32-19, 80-72, 15-90, 42-49, 47-4, 47-102, 47-81, 85-63, 4-102, 77-83, 102-81, 3-29, 30-7, 100-84 and 4-81. Low distances of d = 1 are observed between 6 and 19, 17 and 19, 41 and 40.