About Flagship 1
Plant breeders work to improve the characteristics of plants, especially those related to productivity. The most important productivity traits are yield (the quantity of a given useful product, such as grain) and yield stability (the ability of plants to maintain high yields under stressful conditions).
Most crops are exposed to stresses such as drought or high temperatures, often combinations of several types of stress at the same time. Indeed, this is one of the main challenges associated with climate change. It is relatively simple to select plants for high yield, but much more difficult to select for yield stability because genetic progress is masked by the effect of environmental conditions on the phenotype, a phenomenon known as the genotype x environment (GxE) interaction. The more genetic information we have, the easier it becomes to understand GxE interactions and how to select for yield stability.
Flagship Project 1 is developing genomic and digital solutions to probe the genetic basis of physiological factors affecting yield and yield stability, as well as quality traits (such as seed protein and oil). We have generated specialized structured populations of wheat, canola and lentil that we can test under controlled environments and in the field, and by combining these assets with genotyping datasets, digital phenotyping and data modelling from field environments, we can finally dissect the genetic factors responsible for yield stability under stress. The project will deliver richly-annotated phenotype datasets (phenomics) to support genetic dissection and breeding (genomics), allowing the association of genotypes with phenotypes, ultimately delivering novel genomic and digital phenotyping tools to breeders to support the prediction of crop performance under stress.
The following projects are currently underway within Flagship 1:
Characterization of novel genomic variation in NAM populations
Cataloguing the genomic features of specialized crop populations and develop tools for their visualization.
We have produced structured populations of wheat, canola and lentil based on a technique known as nested association mapping (NAM) and aim to generate a catalogue of genomic information from 2500 recombinant inbred lines in each population as a resource for genotype-phenotype association and the development of selectable markers for trait improvement. We are also developing tools that allow researchers to quickly evaluate and compare the structure of plant genomes, including the positions of genes, sequence variations, and areas of synteny containing similar groups of genes in different crops.
We have extracted high-molecular-weight DNA from the canola NAM population to generate libraries for long-read sequencing, whole-genome assembly, and the identification of structural variation. A full‑genome assembly is now available covering ~90% of the genome. Pipelines have been established to identify genome rearrangements resulting from fixed homoeologous recombination events, and five NAM parental lines have been assessed for such events. The wheat and lentil NAM populations are both being genotyped by exome capture sequencing. We have developed a tool (SynVisio) for the comparison of features and syntenic regions between two or more genomes (or subgenomes in polyploid species such as wheat). This tool also captures system snapshots, displays multiple tracks outside the primary view, and allows additional format options for downloaded images.
Using the Digital Phenotype to Augment Genomic Selection
Collecting multi-environment digital phenotyping data for large structured populations of canola, wheat and lentil for sophisticated genomics and phenomics analysis.
We are using aerial digital phenotyping (the collection of time-series image data from plants using remote sensors) in our canola, wheat and lentil NAM populations to find digital image signatures as proxies for yield, yield stability and quality traits. The integration of high-throughput phenotyping data improves the performance of genomic selection for moderately to highly heritable traits by allowing the precise quantification of phenotypic variation, thus facilitating genetic gain by increasing selection intensity during the early stages of breeding.
We have initially concentrated on the production of canola and wheat seed to allow the replication of field trials, and we have now started collecting aerial imaging data for stitching and plot segmentation to allow for subsequent data analysis. We are testing computational approaches to predict the relationships between genotype, environment and phenotype, including traditional genome-wide association studies as well as novel approaches such as deep learning and probabilistic graphical models. These test cases will allow us to evaluate the performance of different methods in subsequent breeding programs.
Targeted phenotyping for traits impacting yield stability and seed quality
Using novel controlled-environment phenotyping platforms that capture whole-plant physiology, root development and the transport of stress-associated molecules to study the response to water deficit and the impact on yield and seed quality.
Our initial LemnaTec dataset highlighted the limitations of this platform for crops with indeterminate growth (such as canola) but nevertheless allowed the development of segmentation algorithms and machine learning pipelines. The PlantArray system is under construction and should be ready by the end of 2020 in an upgraded greenhouse facility with fully adjustable LEDs. We have prepared a custom leaf chamber for the BioPET to allow the imaging of radiolabelled sugars moving from shoot to roots. We have also shown that a radiolabeled plant hormone (abscisic acid, ABA) produces two derivative metabolites, which we have now synthesized in non-radiolabelled form to use as standards in our future experiments. Radiolabelled ABA and ABA glucose ester standards have also been prepared to investigate ABA metabolism in canola.