Comparisons of populations multiplied for up to two cycles in different environments
Morpho-phenological characterisation and differentiation
The ANOVA analysis showed that the entries differ for all of the morpho-phenological traits recorded with great differences generally observed between the populations and three of the hybrids (HH, HHH, HHHH) (Table 2). For the traits weight of the main head (YLD) and number of secondary heads (HN) the populations derived from the initial synthetic by multiplication for two years were not different from each other. In particular, YLD values ranged from 11.05 g (Syn2-TER) to 324.58 g (HHH), as populations formed a little main head and many (up to 36) secondary heads while most of the hybrids did not form any secondary heads HHH, among the hybrids, and Syn2-FR, among the populations, showed the highest total yield (324.8 and 266.2 g, respectively).
Considering the inflorescence emission and maturation date (DH range = 101 - 204 dd; DM range = 1341 - 219 dd), generally, the latest were the populations, the earliest the hybrids, with the exception of H. LR, Syn1-PG, Syn2-FR, Syn2-UK and Syn2-GR were the latest for inflorescence emission date, with Syn2-GR also late in maturation. Regarding traits related to head characteristics, generally, hybrids showed higher values than populations: HeH range = 6.41 - 9.59 cm, (Syn2-TER and H, respectively), HeW range = 3.29 - 14.53 cm (Syn2-UK and HH, respectively) and SW range = 0.94 - 3.69 cm (Syn2-TER and HHHH, respectively). The hybrids showed the lowest plant height, while LR and Syn1-PG the highest (PH range = 22.05 - 50.78 cm, HHHH and LR, respectively). Regarding plant diameter (PD range =36.31 - 80.39 cm, in H and Syn2-PG, respectively), all the entries showed similar values, with the exception of H, HH and Syn2-TER showing the lowest values. No differences were found among populations regarding vigour with the only exception of Syn2-TER (4.0) which was the less vigorous one.
The first two PCA axes accounted for 83.57% of total data variability, with the addition of the third axis the 97.06% of the total variability was explained (Figure 2). The entries were clustered in two groups: the first one included HH, HHH and HHHH, while the second included the populations with Syn2-TER and H as subgroups of this cluster. Most of the morpho-phenological traits had large effects on the ordination of entries (as indicated by the lengths of their vectors along the first two Axis).
The variables were clustered into four groups: the first one included YLD, SW and HeW, the second one included T-YLD and HeH, the third one PD and V, while the last group included all the other variables. Within each group, variables were highly positively correlated, but the variables of the first group were negatively correlated to those belonging to the third group. The variables of second group were not correlated with those included in the third group, while they were positively correlated with those belonging to the first group (Figure 2).
Genetic characterisation and differentiation
Averages and standard errors relative to the main genetic parameters calculated by entry over the 22 SSR loci are reported in Table 3.
Mean and standard errors relative to the number of successfully analysed genotypes (N), observed (Na), and effective alleles (Ne), observed heterozygosity (Ho), expected heterozygosity (He) and Fixation Index (F) worked out for original LR, Syn1-PG, its derived populations by two years of multiplication across Europe and hybrid controls on the basis of data recorded overall for the 22 markers used.
All the entries were characterised by similar number of alleles (N). In general, Na was higher than Ne in all the entries. In populations, Na ranged from 2.45 (±0.205, Syn2-TER) to 3.50 (±0.327, LR) and Ne ranged from 1.78 (±0.126, Syn2-FR) to 2.38 (±0.188, Syn1-PG), hybrids generally had lower values of Na and Ne than the populations.
Populations did not differ for Ho and He, while obviously differed from hybrids.
Fixation index (F) values were close to 0 for all the populations. The existence of negative values in some populations and in all the hybrids hinted to a heterozygosity excess.
Finally, all Fst values obtained from pairwise comparisons were highly significant (P<0.01), with the only exception of the comparison between LR and Syn1-PG (Table 4). Considering two years of multiplication, significant pairwise Fst comparisons between the LR and derived populations increased from 0.026 to 0.066. In addition, a crescendo of differentiation was found between Syn1-PG and populations derived by multiplication in other environments (from 0.039 between Syn1-PG and Syn2-GR, to 0.070 between Syn1-PG and Syn2-TER).
Similarly, a crescendo of differentiation was found between Syn2-PG and the same generation of multiplication in different environments (from 0.046 between Syn2-PG and Syn2-GR to 0.103 between Syn2-PG and Syn2-TER). The highest differentiation value was found between Syn2-FR and Syn2-TER, populations which were multiplied in very contrasting environments. As expected, the genetic differentiation between hybrids and between populations and hybrids was very high.
AMOVA (not shown) worked out on the total data showed that among- and within- population diversity accounted for 33% and 67% of total genetic variation, respectively. However, when it was worked out only considering the LR and its derived populations, the proportions were 11% and 82%, respectively.
NJ tree results showed the presence of two main clusters (Additional file 4: Figure S3). A cluster included most of the individuals belonging to populations while the other one to the hybrids. The first cluster showed others, although not completely differentiated, sub-clusters: Syn2-TER, Syn2-FR and Syn2-UK genotypes mostly clustered together. The within-population genetic variation was evident in populations, while hybrids generally showed a high genetic uniformity in accordance with their genetic structure.
The first three components of the PCoA accounted for 85.90% of genetic variation (Figure 3). PCo1 (52.94%) clearly separated populations from hybrids, while PCo2 (26.47%) separated the hybrid H from the other entries. PCo3 (6.49%) mainly discriminated Syn2-FR and Syn2-TER from Syn2-UK and Syn2-GR.
Comparison of populations multiplied for up to three cycles in Central Italy
Morpho-phenological and genetic characterisation
The ANOVA analysis showed that the assessed entries differ for many of the morpho-phenological traits recorded (Additional File 5: Table S2). All populations were clearly discriminated from HH, HHH and HHHH. However, populations multiplied at the same site, with very few exceptions, did not show much differences and, generally, all populations obtained after three cycles of multiplication showed similar values to those found in the populations obtained after two cycles of multiplication. Worth of note is only that Syn2-GR was different from Syn3-GR and Syn2-TER from Syn3-TER for HeH, Syn2-PG was different from the LR and the other populations multiplied in Perugia for DH, DM and PD and, finally, Syn3-PG was different from LR, Syn1-PG and Syn2-PG for V.
The PCA (with the first three axes accounting for 96.54% of total data variation) and the biplot showed that the three hybrids and hybrid H with populations clustered separately and that the variable clustering and relationship were similar to what previous observed (i.e. considering the morpho-phenological comparisons of populations multiplied for up two cycles in different environments, Additional File 6: Figure S4)
Averages and standard errors relative to the main genetic parameters calculated by entry over the 22 SSR loci are reported in Additional File 7: Table S3. A part differences between entries, population obtained with three years of multiplication did not significantly differ from those obtained with two years of multiplication for all genetic parameters, with the exception of Na observed in Syn2-GR and Syn3-GR. Finally, pairwise Fst comparisons (Table 4) ranged from 0.005 to 0.159 between populations. Population differentiation increased across locations and generations of multiplication. The highest differentiation was found between Syn2-TER and Syn3-TER.
NJ tree results, beside a clear differentiation between populations and hybrids, showed that Syn2-TER, Syn3-TER, Syn3-GR, and partially Syn3-PG, appeared to belong to four different and distinguished clusters (Additional File 8: Figure S5). The first three components of the PCoA accounted for 82.63% of genetic variation (Additional File 9: Figure S6). Populations and H were clearly discriminated from HH, HHH and HHHH. Finally, Syn3-GR, Syn2-TER and Syn3-TER were clearly differentiated from all the other populations.