Targeted generation of polyploids in Hydrangea macrophylla through cross-based breeding.

BACKGROUND: Up to now, diploid and triploid cultivars were reported for the ornamental crop Hydrangea macrophylla. Especially, the origin of triploids and their crossing behaviors are unknown, but the underlying mechanisms are highly relevant for breeding polyploids. RESULTS: By screening a cultivar collection, we identified diploid, triploid, tetraploid and even aneuploid H. macrophylla varieties. The pollen viability of triploids and tetraploids was comparable to that of diploids. Systematic crosses with these cultivars resulted in viable diploid, triploid, tetraploid and aneuploid offspring. Interestingly, crosses between diploids produced diploid and 0 or 1-94% triploid offspring, depending on the cultivars used as pollen parent. This finding suggests that specific diploids form unreduced pollen, either at low or high frequencies. In contrast, crosses of triploids with diploids or tetraploids produced many viable aneuploids, whose 2C DNA contents ranged between the parental 2C values. As expected, crosses between diploid and tetraploid individuals generated triploid offspring. Putative tetraploid plants were obtained at low frequencies in crosses between diploids and in interploid crosses of triploids with either diploid or tetraploid plants. The analysis of offspring populations indicated the production of 1n = 2x gametes for tetraploid plants, whereas triploids produced obviously reduced, aneuploid gametes with chromosome numbers ranging between haploid and diploid level. While euploid offspring grew normally, aneuploid plants showed mostly an abnormal development and a huge phenotypic variation within offspring populations, most likely due to the variation in chromosome numbers. Subsequent crosses with putative diploid, triploid and aneuploid offspring plants from interploid crosses resulted in viable offspring and germination rates ranging from 21 to 100%. CONCLUSIONS: The existence of diploids that form unreduced pollen and of tetraploids allows the targeted breeding of polyploid H. macrophylla. Different ploidy levels can be addressed by combining the appropriate crossing partners. In contrast to artificial polyploidization, cross-based polyploidization is easy, cheap and results in genetically variable offspring that allows the direct selection of more robust and stress tolerant polyploid varieties. Furthermore, the generation of polyploid H. macrophylla plants will favor interspecific breeding programs within the genus Hydrangea.

Rapid identification of inflorescence type markers by genotyping-by-sequencing of diploid and triploid F1 plants of Hydrangea macrophylla.

BACKGROUND: The ornamental crop Hydrangea macrophylla develops highly attractive lacecap (wild type) or mophead inflorescences. The mophead trait, which is mostly favored by consumers, is recessively inherited by the INFLORESCENCE TYPE locus (INF). If lacecap cultivars are crossed with mophead cultivars, then either 50% or all progenies develop lacecap inflorescences, depending on the zygosity at the INF locus. For most cultivars, the zygosity at the INF locus is unknown. Furthermore, the determination of the inflorescence type in offspring populations is time-consuming, because seedlings flower the first time in the 2nd year after sowing. Within this study, we aimed to develop DNA-based markers that allow to determine the zygosity at the INF locus of prospective parental plants and to predict the inflorescence phenotype of seedlings already in the non-flowering stage. RESULTS: By crossing a mophead and a lacecap cultivar of H. macrophylla, we produced a pseudo-backcross F1 population consisting of 422 plants. These plants segregated into 279 lacecap, 73 mophead, 3 intermediate and 67 non-flowering plants, differing significantly from the expected 1:1 segregation ratio. Surprisingly, 75% of these plants were triploid, although both parents were diploid. We found that the lacecap parent produced unreduced pollen, which induced the formation of triploids. 380 randomly selected F1 plants were genotyped by genotyping-by-sequencing (GbS). Using a genome assembly of cultivar 'Sir Joseph Banks', we performed subsequently a bulk sequence analysis with pooled GbS data of diploid versus mophead plants. We identified directly 2 markers tightly linked with the INF locus, each of them explaining 99.7% of the inflorescence phenotype. Using a collection consisting of 56 diploid, triploid or tetraploid H. macrophylla varieties, we detected 6 sequence variants for one of these markers. Two variants were associated with the mophead phenotype. Furthermore, we found by marker analysis a co-segregation between the mophead and the non-flowering trait, which indicates a major flowering time locus next to the INF locus. CONCLUSION: Through bulk sequence analysis of pooled GbS data from diploid and polyploid F1 plants, we identify rapidly tightly linked markers for the inflorescence type, a dominant-recessively inherited trait in the non-model plant species H. macrophylla.