Towards marker assisted breeding in garden roses: from marker development to QTL detection

Over the last few decades the rose market in Eastern Europe showed a steady growth, which indicates that there is increasing demand for new cultivars that are adapted to the climate as well as to the customs and beauty criterion of that region. One of the possibilities to speed up breeding is to implement marker assisted selection (MAS). Implementation of MAS requires a specific infrastructure (molecular markers, knowledge on genetics of important traits, genetic maps) which is not yet available for tetraploid roses. In this thesis I developed some of the prerequisites for MAS in roses and discuss when and how MAS could have a positive effect on accelerating breeding and/or reducing the costs of the breeding process. The first step in understanding the structure of the genepool of garden roses was to evaluate the relatedness among available cultivars. For the first time genetic diversity among modern garden rose cultivars was evaluated (<strong>Chapter 2)</strong> using a set of 24 microsatellite markers covering most chromosomes. A total of 518 different alleles were obtained in a set of 138 rose cultivars. Genetic differentiation among types of garden roses (Fst=0.022) was four times that found among cut roses, and similar in magnitude to the differentiation among breeders, due to the fact that horticultural groups and breeders overlap largely in classification. In terms of genetic diversity cut roses can be considered as a subgroup of the garden roses. Winter hardy Canadian garden rose cultivars (Explorer roses) showed the least similarities to European roses, and introgression from wild species for winter hardiness was clearly visible. Roses of two breeding programmes (Harkness and Olesen) shared a similar genepool. Comparison of the differentiation among linkage groups indicated that linkage group 5 is potentially a region containing important QTLs for winter hardiness. Linkage group 6 contains the largest amount of genetic diversity, while linkage group 2 is the most differentiated among types of garden roses. Garden roses, as well as many other important crops (wheat, potato, strawberry, etc.) are polyploid. Genetic analyses of polyploids is complex as the same locus is present on multiple homologous chromosomes. SSR markers are suitable for mapping in segregating populations of polyploids as they are multi-allelic, making it possible to detect different alleles of the same locus on all homologous chromosomes. If a SSR marker gives fewer alleles than the ploidy level, quantification of allele dosages increases the information content. In <strong>Chapter 3</strong> I showed the power of this approach. Alleles were scored quantitatively using the area under the peaks in ABI electropherograms, and allele dosages were inferred based on the ratios between the peak areas for two alleles in reference cases in which these two alleles occurred together. We resolved the full progeny genotypes, generated more data and mapped markers more accurately, including markers with “null” alleles. Even though SSR markers are one of the most appropriate marker systems for genetic studies in polyploids still few hurdles complicate (reduce) their implementation. The first major hurdle in developing microsatellite markers, the cloning step, has been overcome by next generation sequencing techniques. The second hurdle is the testing step to differentiate polymorphic from non-polymorphic loci. The third hurdle, somewhat hidden, is that only those polymorphic markers that detect a large effective number of alleles in the germplasm to be studied, are sufficiently informative to be deployed in multiple studies. Both selection steps are laborious and still done manually. In <strong>Chapter 4</strong> I present a strategy in which we first screen sequence reads from multiple genotypes for repeats that show the most variation in length, and only these are subsequently developed into markers. We validated our strategy in tetraploid garden rose using Illumina paired-end transcriptome sequences of 11 roses. Out of 48 tested two markers did not amplify but all others were polymorphic. Ten loci amplified more than one locus, indicating duplicated genes or gene families. Completely avoiding this will be difficult, as the range of numbers of predicted alleles of highly polymorphic single- and multi-locus markers largely overlapped. Of the remainder, half were duplicates, indicating the difficulty of correctly filtering short sequence reads containing repeat sequences. The remaining 18 markers were all highly polymorphic, amplifying between 6 and 20 alleles in the 11 tetraploid garden roses. This strategy therefore represents a major step forward in the development of highly polymorphic microsatellite markers. Despite that garden roses are economically very important ornamentals, breeding is still mostly conventional, mainly due to tetraploidy and the lack of genetic maps and knowledge about the genetic base of important traits. Furthermore, crosses with unintended parents occur regularly and detection of these is not always straightforward, especially when genetically related varieties are used. Moreover, in polyploids detection of off-type offspring often relies on detecting differences in allele dosage rather than the presence of new alleles. In <strong>Chapter 5</strong> I applied the WagRhSNP Axiom rose SNP array to generate 10,000s of SNPs for parentage analysis and to generate a dense genetic map in tetraploid rose. I described a method to separate progeny into putative populations which share parents, even if one of the parents is unknown, using PCO analysis and sets of markers for which allele dosages are incompatible. Subsequently, dense SNP maps were generated for a biparental and a self-pollinated mapping population with one parent in common. I confirmed a tetrasomic mode of inheritance for these crosses and created a starting point for implementation of marker-assisted breeding in garden roses by QTL analysis for important morphological traits (recurrent blooming and prickle shape). Winter hardiness is a complex trait and one of the most important limiting factors for garden rose growth and distribution in areas characterized by a continental climate. In <strong>Chapter 6</strong> research was undertaken to determine the genetic regions underlying winter hardiness of garden roses, and to generate markers linked to them. For this purpose we exposed two segregating populations, RNDxRND and RNDxHP, to temperatures below -15C in a cold chamber and in the field in Serbia. The frost damage in the hardened plants was estimated directly at the phenotypic level (proportion of dieback) and at the non-visible physiological level indirectly (through the potential for meristem production in spring; regrowth). For winter hardiness we detected two tentative QTLs in the RNDxRND population and two tentative QTLs in the RNDxHP population, of which one was the same in both populations. The ability of plants to regrow in spring was associated to genomic regions on three linkage groups of the RNDxRND population, and on two different linkage groups in the RNDxHP population. A comparison of the ability for regrowth and level of damage caused by low temperature revealed that these two traits are inherited independently and that the final cold tolerance depends on the plant’s ability to withstand low temperature and to regrow fast in spring. In résumé, this thesis resulted in the development of basic tools (a fast strategy for polymorphic SSR marker development), basic methods/concepts for genetic analyses in polyploids (quantification of SSR allele dosage, distinguishing outliers from population in polyploid crops, dense SNP map generation and QTL study in tetraploids), and knowledge on genetics of important traits in rose (relatedness among modern garden roses (genetic diversity approach), mode of inheritance, occurrence of selfing, QTLs for morphological traits (recurrent blooming and prickle shape) and dissection of winter hardiness (level of damage caused by low temperature and regrowth)). Additionally, potential use of markers in every phase of rose breeding was discussed (<strong>Chapter 7</strong>). All these aspects contribute to a solid basis for marker assisted breeding in (garden) rose.