The allo-octoploid strawberry: simply complex

The garden strawberry (Fragaria x ananassa) is a fruit species that was developed through human intervention less than 300 years ago. Currently, it is the most important soft fruit in both production as well as value and renowned for its deliciousness. There are many challenges in growing such a delicate fruit, many of which have been overcome through improved cultivation techniques and breeding. The perishability of the product is, however, not the only challenge faced by strawberry breeders. In terms of genome composition, strawberry appears to have accumulated a wonderful array of obstacles to genetic studies. It is a vegetatively propagated allo-octoploid outbreeder, and only few crop species are worse of in this respect. Many of the molecular genetic ground work is therefore performed in its diploid ancestor, the woodland strawberry Fragaria vesca, which was sequenced in 2011. However, since nearly all strawberries that are eaten are octoploid, genetic research can’t linger at the wild diploids forever. In this thesis we developed new tools and analysis methods for genetic studies in the allo-octoploid strawberry and subsequently applied these methods in the detection of marker-trait associations.

The purpose of Chapter 2 was to develop a method to interpret the complex peak patterns generated by microsatellites in octoploid strawberry in such a way that we ended up with as much information as one would expect to retrieve from a microsatellite in a diploid system. This information could then be used to generate high quality linkage maps for the different sub-genomes and allow for easy alignment and comparison. We named the method MADCE, which stands for Microsatellite Allele Dose & Configuration Establishment. In the MADCE methodology, we first need to determine the dose of each allele present in an individual. For this we used the signal of fluorescent microsatellite peaks in relation to the total fluorescent signal generated by all peaks for that microsatellite. We then used the disomic inheritance of strawberry to establish the allelic configuration of each different homoeologue (subgenome). The repulsion of alleles from the same subgenome in offspring allowed us to form subgenomic pairs of alleles. We found that in single cross mapping populations, the deployment of our method was fairly easy due to the high number of offspring that can be used to establish repulsion between alleles. However, for pedigreed breeding germplasm this was another matter, as generally only few offspring were available. For this we added some additional tricks to the MADCE method, although some uncertainty about the configuration would remain for problematic lines and alleles.

In Chapter 3 we used the MADCE method from Chapter 2 to generate a genome wide linkage map for the Holiday x Korona (HxK) mapping population. This linkage map was to be used in subsequent experiments for QTL discovery as well as provide the strawberry community with a highly detailed map consisting not only of marker distances, but allele and haplotype configuration of the parents Holiday and Korona as well. The haplotype information revealed that inbreeding (homozygosity) levels in Holiday were similar to the levels expected from its pedigree, but that inbreeding levels of Korona were more than three times higher than expected, which could be resultant from selection pressure enacted by breeders. Selection pressure could also be causal to our discovery that the kinship between the two cultivars was twice as high as expected from their shared ancestry. Another discovery was a large inversion on one of the subgenomes of linkage group 2 (D). Up until the publication of our linkage map this inversion had not been reported in other linkage maps. Another innovation was our attempt at giving a biological or evolutionary meaning to the denomination of the linkage groups by arranging them according to similarity to the diploid ancestor F. vesca, based on F. vesca derived primer amplification efficiencies. The HxK map has been used in several (ongoing) research projects outside of our research group and has contributed to the development of the 90k Axiom SNP array for cultivated strawberry.

In Chapter 4 we performed a QTL mapping study for disease resistance against the problematic pathogen Phytophthora cactorum, which causes crown rot in strawberry plants. In this study we used two different mapping populations: the Holiday x Korona (HxK) population from the previous chapter as well as E1998-142 x Elsanta (ExEls), developed more specifically for the purpose of finding resistance against P. cactorum. The HxK and ExEls populations were phenotyped over three years (2008, 2010 and 2011) under different seasons and conditions. The correlation between years for was quite low for both populations (ranging from 0.18 to 0.47), indicating a large environmental effect on disease pressure. Results from the QTL analysis showed that most QTLs were small in effect and only just above the statistical significance threshold. Only for ExEls we uncovered two QTLs with relatively high significance levels, but none were significant in all three years. Because of the high environmental influence, and the desire to have QTLs that are robust over environments, we used the average of all three years (AOTY) as an additional phenotype. When we used the AOTY trait, the QTL on LG7D became stronger than for any of the individual years. Whereas for the LG7C QTL the significance dropped to just below threshold levels. These results indicated that removing environmental noise through averaging over experiments is a good way to uncover the most reliable and therefore more valuable to a breeding program.

In Chapter 5 we investigated the genetics behind two different flowering habits that are grown commercially worldwide: seasonal flowering habit (SF) and perpetual flowering (PF) These varieties initiate flowering under long days, and can therefore produce fruit for a much longer period: throughout the summer and early fall. Evidence from literature and practical breeding suggested that PF is under dominant control. We decided to treat PF as a qualitative trait and divided two small mapping populations into PF and SF individuals. After screening several microsatellites, we found one locus that completely cosegregated with the PF trait at the bottom of LG4D. At the moment of mapping, a paper was published which mapped the same trait to the same location. We found that there were two very clear candidate genes within our QTL interval, FaCDF2 and FaFT2, which were homologous to genes that are major factors in the flowering pathway of Arabidopsis and many other plant species. We then sequenced the FaCDF2 gene from a number of distinct PF and SF cultivars. This resulted in the discovery of two quite distinct allelic variants, one of which was present in all PF cultivars. However this variant was also present in some of the SF cultivars, indicating that either FaCDF2 is not the causal gene, or that other loci can have a qualitative effect on the switch from SF to PF. We then performed microsatellite haplotyping on hundreds of cultivars and this revealed that all PF varieties of all origins carry the same haplotype in the PF QTL region, and that there weren’t any recombinations between the candidate genes FaCDF2 and FaFT2, which are 250kb apart on the physical genome. This makes it still undecided which of these two candidate genes are causal to the PF trait. Another interesting result from the haplotyping was that the complete PF haplotype was present with moderate frequency in SF varieties as well. Not only does this suggest a common origin, it also complicates the establishment of a theory for the mechanisms behind perpetual flowering in cultivated strawberry. So far we have not been able to establish whether the PF haplotype that is present in SF cultivars is functionally distinct from the PF haplotype in PF cultivars. All we know is that it does not confer perpetual flowering in these SF cultivars, and further experiments would be needed to find out the exact mechanism behind perpetual flowering.

In the general discussion (Chapter 6), the results of this thesis were placed in the broader context of science in general and plant breeding in particular.