Marco Sandri, Paola Zuccolotto
Giovanni Battista Tornielli, Marianna Fasoli, Sara Zenoni
The Molecular Phenology Scale (MPhS) is a tool aimed to map the ontogenetic development of the fruit.
The statistical pipeline developed to define the MPhS is an unsupervised learning procedure yielding an innovative combination of semiparametric, smoothing and dimensionality reduction tools. From a statistical perspective, the most noticeable features of the proposed method are that (1) time information is exploited by considering only the ordering of timepoints and not their distance and (2) the Principal Components extracted at Step 3. are exceptionally able to summarize different characteristics of data (berry variety, vintage, stage). From issue (1) follows that the MPhS measurement unit is not time, but an ideal step of the berry development process, which can take a longer or a shorter time, according to a multiplicity of different factors. From issue (2) follows that we have been able to select only genes directly involved with the berry development process, depurating from those impacted by the berry variety and the vintage effects.
The proposed scale paves the way for the development of tools that aspire at predicting the phenological stage of the fruit in various climate conditions, such as models that can account for temperature and other environment clues. The quality of these tools will benefit from the combination of various modeling techniques (molecular, metabolite, physical, visual levels), providing that great coordination and knowledge-transfer between modelers, biologists and growers will be established.
The proposed pipeline could be potentially extended and successfully applied to any other fruit species, provided they have some basic requirements:
Giovanni Battista Tornielli, Marco Sandri, Marianna Fasoli, Alessandra Amato, Mario Pezzotti, Paola Zuccolotto, Sara Zenoni (2022), A molecular phenology scale of fruit development.
Fruit growth and development is defined by phenological scales that deem descriptors such as visual/physical traits or easy-to-measure compositional parameters. The precise identification of a fruit growth stage may be hindered by seasonal variation that is especially relevant in perennial crops grown in the field like the grapevine. In this work molecular-based information from several grape berry transcriptomic datasets was accessed to build a molecular phenology scale and to map the ontogenetic development of the fruit. A portion of the transcriptome exhibiting conserved expression dynamics throughout fruit development across genotypes and growing conditions was selected. The proposed statistical pipeline consisted in an unsupervised learning procedure yielding an innovative combination of semiparametric, smoothing and dimensionality reduction tools. The molecular scale allowed the alignment of time-series fruit samples and proved to be a step forward in mapping the progression of fruit development with higher precision compared to classic time- or phenotype-based approaches.
Supplementary material
Giovanni Battista Tornielli, Marco Sandri, Marianna Fasoli, Nick Dokoozlian, Mario Pezzotti, Paola Zuccolotto, Sara Zenoni, A novel berry phenological scale based on gene expression, 11th International Symposium on Grapevine Physiology & Biotechnology, 31 October – 5 November 2021, Stellenbosch, South Africa.
Phenology scale systems widely adopted by viticulturists define stages of the annual development of the vine based on the visual description of well recognizable traits related to organ growth and morphology, including grape features from fruit set to maturity. However, although some stages can be easily described (e.g. fruit set, veraison), defining a comparable developmental stage for grapes of the same cultivar when grown in different conditions or for grapes of different cultivars can be challenging, in particular after the onset of ripening. By analysing transcriptomic data collected over berry development, it was shown that the variations of a portion of the transcriptome exhibited conserved dynamics across cultivars and growing condition of grapevines, and thus may be used to describe the developmental stage of berry development. In this work, we used the transcriptomic data generated from grape berries weekly sampled from Cabernet Sauvignon and Pinot noir vines grown in the same location over three consecutive vintages, focusing on conserved annual dynamics rather than on the biological significance of the expression program inferred by gene function. By interpolating the transcriptomic samples dispersed in a 3D space of a PCA we built a 30-stage Transcriptional Phenology Scale (TPhS) precisely defining the progression of development from berry formation to full ripening. The performance of the scale was assessed projecting onto the TPhS both RNA-seq and microarray transcriptomic samples from the same dataset used to elaborate the scale, and from several other public datasets. The results allowed to align samples on the new phenological scale and to highlight differences related to variables like the grape variety, the cultivation site, the vintage, or the applied treatment such as cluster thinning, defoliation, water limitation and temperature regimes. In some cases, the phenological re-scaling of sample collections from previous studies provided valuable hints to re-interpret the experimental results. Overall, we show that the transcriptomic information can be accessed to precisely define a transcriptional phenology scale that can be used to map the ontogenetic development of the fruit with high precision and to align the stage of berry development of different grapes.
