Insights into the genetic control of plant architecture in apple

Apple (Malus x domestica) is one of the most important fruit crops in the world and growers are continuously seeking to improve fruit production and fruit quality while reducing management costs. Yield, homogeneity of fruit quality and the time needed for young trees to reach full production are influenced by the tree’s architecture. Tree architecture is currently controlled through orchard management practices, but these practices can be time-consuming, labor-intensive and expensive.

A wide variety of tree architectures exist in apple. However, little is known on tree architecture and its genetic control in apple and most studies have focused on limited genetic backgrounds. To increase the knowledge tree architecture in apple, the aims of this thesis were (1) to establish the tree architecture of young trees in a collection of 130 genotypes, both local accessions and commercial varieties, supplemented with 25 F1 individuals of a cross between ‘Telamon’ and ‘Braeburn’ and (2) identify genomic regions that control tree architectural traits using genotypic information from the IRSC 8K apple SNP array.

The formation of sylleptic branches is desired during the first year of growth. Unfortunately, very few of these branches were formed in the first year. Up to four trees per accession were headed to stimulate sylleptic branching. The heading significantly increased the amount of syllepsis in the subsequent year although not all genotypes reacted equally. Generally, genotypes that grew more vigorously and formed more branches before heading, showed a stronger reaction to heading. Thus, heading can be advantageous when limited branches were formed and when cultivars are known to respond positive to the heading. No correlation between the response to heading and the genetic substructure of the germplasm was observed.

Apple trees have been classified into growing types based on overall tree shape and/or fruiting type. Recording the growth of one-year-old and two-year-old trees showed that most architectural traits are genetically controlled, making selection for these traits possible. No correlation was found between individual component traits of different branching types, suggesting that they could be controlled and selected for independently. However, cluster analyses did not show a clear clustering of different growth types, indicating that a ‘tree architectural continuum’ exists rather than well-defined growing types.

The IRSC 8K apple SNP array was used to genetically characterize the used germplasm and F1 individuals. Unfortunately, the whole-genome duplication event in apple, other duplicated segments, the sub-optimal reference genome and high heterozygosity of apple have led to non-specificity or binding problems to the probes on the array and issues with the automatic genotype scoring algorithm. The suggested parameters to remove SNPs with erroneous genotype calls were not sufficient and additional parameters and thresholds were identified to choose SNPs with reliable genotype calls.

The obtained set of 2095 SNPs was used to genetically characterize the germplasm collection. The genetic characterization identified triploids and genetically identical individuals that had to be removed for the association anlyses. Most known pedigree records could be confirmed and new putative parent-child relations for non-commercial accessions were identified. Two major subgroups were identified: one mainly consisting of the local, non-commercial varieties and a second group mainly consisting of commercial cultivars. A rapid decline of linkage disequilibrium was observed within 80-100 kb which indicated that more SNPs are needed to cover the whole genome during association analyses.

An association analyses resulted in associated SNPs for nine traits while no associations were found for the other seventeen traits. Although almost half of the genes near identified SNPs had an unknown function, multiple candidate genes were identified based on existing literature, such as orthologues of MAX4, a biosynthesis gene for a plant hormone that inhibits branching, and BRC1, an important integrator of multiple branching-regulating signals. These genes are excellent targets for further research.

In conclusion, the results of this study indicate that differences in growth and growth reactions to manipulations exist between accessions in apple and that these differences are partly due to the genetics of these accessions. This means that orchard practices should be adjusted to the cultivars but also that breeding for apple genotypes with specific growth characteristics is possible. However, associated loci could not always be found. Further studies with higher density arrays are needed to identify loci controlling these traits or to indicate that no large-effect loci exist and that these traits are controlled by many small-effect loci. In the latter case, genome-wide selection and the establishment of breeding values for architectural traits could be the option when breeding for tree architecture in apple.