T-Lidar as a new high-throughput methodology for studying the genetic determinisms of apple tree architecture

International audience

Английский. Growth, branching and reproduction processes are highly organized in space and time in plants. In many Rosaceae fruit tree species, this organization has been described by architectural analyses to evaluate tree plasticity in response to environmental conditions and genetically determined traits. However the study of genetic determinism of architectural traits requires the observation of large populations of individuals over several years which is hardly compatible with manual or classical digitizing techniques. We thus investigated new solutions based on T-Lidar technology that is currently developed mainly to characterize forest stands. First, we explored the possibility to extract from T-Lidar scans features of tree form on a limited set of genotypes with contrasted architectures. Different precisions and distances of laser scans were tested to assess sensitivity of branches detection and determine an optimal combination between acquisition times and scan resolution. The scan analysis required the use of different algorithms to segment and extract the point clouds corresponding to each individual tree and to rebuild their 3D architectures. This allowed us to extract several descriptors of tree height and shape, total leaf area, leaf area distribution and density and light interception efficiency. Tree reconstructions gave promising results with R² ranging from 0.91 to 0.78 for the total number of axes and leaf area when T-Lidar data were compared to digitizing or manual measures. Second, scans were collected on a large range of individuals (around 1000) corresponding to the French core collection of apple varieties. All variables showed a high variation and heritability and architectural morphotypes were distinguished within the population by a clustering method. GWAS analyses are currently performed. The next steps of our research will be the creation of a dedicated workflow for scan analysis allowing faster computations and the improvement of the set of descriptors related to tree topology that could be used for genetic studies.