Heterosis-relevant QTL were detected for numerous traits in field and greenhouse trials.
By combining data from the DH and BC populations, classes of QTLs were determined based on their effects. How strongly each QTL affected a particular trait was important in associating patterns of phenotypic level with the QTL that were detected. One may assume, for instance, that two genetically correlated traits, as estimated in this study, can be traced back to the same QTL. Alternatively, a QTL for a trait may affect another QTL responsible for another trait, which may lead to a causal relationship. However, if the contribution of a QTL to the trait variation is weak, an attempt to relate QTL to a quantitatively variable phenotype becomes challenging.
The following discussion begins with an interpretation of the phenotypic patterns shown in the respective greenhouse and field trials, followed by an interpretation of detected QTL with a focus on QTL related to heterosis-relevant effects (dominance and dominance-derived epistasis). Finally an attempt will be made to relate the phenotypic patterns related to heterosis to the QTL composition. The results will be discussed in relation to other relevant work and with an outlook towards further investigations of the genetic control of heterosis.
From a quantitative genetics perspective, a population that consists of heterozygous individuals (e.g. BC test hybrids), has an additional driving effect due to dominance and dominance-related epistasis. The presence of these effects alters the additive contribution from each allele due to the interaction between them (additive-to-additive interaction).
Although a negative effect is possible, in general heterozygosity brings an overall positive improvement to the phenotypic performance. This is why heterosis and hybrid vigor are often used as interchangeable terms. The results obtained from the present study simply support this expectations, since there were more positive improvements from the DH lines to the respective BC test hybrids.
Another common feature seen in the phenotypic observations of the DH and BC populations is that variability was generally higher among the DH lines than among the BC hybrids. This was the case for almost all of the traits observed in both the greenhouse experiment and the field trials. Such relative performance is expected, simply because the BC hybrids are obtained by backcrossing each DH line with „MSL-Express‟. Thus each member of the BC population shares half of its alleles, meaning that the allelic pool of the BC population has lower variability than that of the DH population.
The field trials over two growing seasons made it possible to compare the performance of the materials between 2005/06 and 2006/07. The DH population performed relatively consistently for plant height, yield and TSM (despite a clearly lower yield and plant height in 2006/07 in comparison to 2005/06), whereas the BC population tended to cluster resulting in a lack of correlation patterns. Mid-parent heterosis had different patterns for each of the traits reported here. Plant height, which was measured at the end of flowering season, had positive heterosis recorded for both years in almost all accession pairs involved in the trial, although an apparent lack of consistency was shown by many of the accession pairs since a low correlation existed between the two years of observations. Yield, on the other hand, showed a clear difference in pattern between 2005/06 and 2006/07. There was clearly more positive mid-parent heterosis in 2006/07 than 2005/06. In 2005/06, roughly half of the DH-BC pairs showed negative MPH. On the other hand, the parental line „Express 617‟ performed poorly in 2006/07, leading to a reduction of mid-parent values and hence a higher estimate of mid-parent heterosis.
4.1.2 Correlations among traits
One aim of this study was to search for possible relationships between early morphological traits with traits in the later stage of development, and particularly with the seed yield. Such information could help us to learn about certain physiological or developmental advantages that are maintained from an early stage throughout the plant‟s lifecycle. Practically, this knowledge may help us to identify gene loci influencing early development that are critical for improved seed yield.
The trait correlation patterns within the mid-parent heterosis dataset did not meet the expectations. There were significant correlations of certain early morphological traits to plant height and yield (Figure 3.14). Shoot dry weight showed a low but signficant correlation to yield in both 2005/06 and 2006/07. Another early trait, specific leaf weight, showed a significant but low correlation to plant height in 2005/06 and yield in 2006/07.
Although low, these correlations indicated that certain relationships among traits existed in the heterotic expression at a morphological level. The DH population showed slightly higher correlations of fresh and dry shoot weight with yield and plant height from both harvest years. These indicated that per se biomass building activity, which estimates additive action, influences different traits separated in time. Thus, we have clear indications for correlations between early and later developmental traits, and these correlations are influenced by both additive and dominance actions.
Quijada et al. (2006) observed correlations among field traits from DH lines and their respective test crosses and concluded that the correlation pattern between traits in the DH and testcross populations were different. Exceptions were between plant height and day to flowering, as well as between seed yield and test weight. Unfortunately, that study provided no information relating early development traits with traits at later development or harvest. In the study of Quijada (2006) there was an indication that yield and plant height had a close additive correlation, based on DH data set analysis, however seed weight was correlated with neither of those traits.
Comparing scatter plots from the 2005/06 and 2006/07 field trials in the present study revealed that heterosis expression in the BC test hybrids was not stable in different environments relative to the performance of the DH lines. That did not mean, however,
that the BC test hybrids had a lower performance. The hybrids still showed higher average trait values than the lines, however the range of trait values was smaller than that of the lines. In other words the performance of the poorest DH lines was compensated by hybridisation, whereas the performance of the best DH lines was not improved so dramatically. This suggests a role of dominance effects that were already expressed in the best DH lines because positive alleles from the recurrent parent „Express 617‟ were already present.