Using models to optimize planting density

You can listen to this page as an audio file.

What is the best way to maximize your harvest: crowding more plants into your plot or spacing them out to maximize light exposure? Computer modelling shows that some plants can modify their leaf architecture to improve light and can therefore tolerate moderate crowding.

For growers of annuals, optimal plant spacing ranging from seed packet suggestions to intensive planting can be determined from a single growing season. For perennial crops such as oil palm, which begin to produce 3 years after planting and continue to produce for 25 years, answering this question would require long and expensive agronomic trials, making the testing of innovative planting practices impractical.

A new study by CIRAD researcher Dr. Raphaël Perez and colleagues tested the impact of planting design and architectural plasticity on physiological responses such as light interception and carbon assimilation.

To determine the extent of architectural plasticity with planting densities, the authors estimated plant biomass and dimensions from harvested 6-year old palms grown under different planting densities. They also conducted LiDAR scans to produce 3D architectural data. “Field measurements of leaf architecture, mainly leaf angles and 3D coordinates, were hardly feasible on a high number of plants and may be very sensitive to manipulators. We found that LiDAR-based measurements are promising for fast and accurate phenotyping of oil palm architecture,” explains Perez.

This data proved that individual palm plants express plasticity. Increased plant proximity increased rachis length and leaf erectness, and decreased leaf weight, whereas other structural traits remained unchanged.

The panel on the left shows a palm tree as the 3D coordinates from the Lidar point cloud. A photograph pointing to the location on the point cloud image shows a target ball stuck to the palm frond. This target is evident on the point cloud image as a dot. Another panel zooms in on the point cloud image to show the detail for one palm frond with points along the rachis highlighted. This leads to an image with just the angle of the rachis. A final panel is an image of the Lidar equipment mounted on a tripod facing a palm tree. — Processing LiDAR scans using PlantScan3D software to retrieve 3D coordinates along the rachis.

An existing oil palm simulation model (VPalm) was then used in combination with a biophysical model (Archimed-φ) to explore how these changes in plant architecture affected light interception and therefore carbon acquisition.

The model generated 3D mock-ups using the field-measured data. The authors were then able to simulate the amount of light intercepted for plants grown at different densities. From this, carbon assimilation at the plot scale could be simulated.

5 palm trees, each representing a different planting design all show different plant architecture. Differences in light interception is evident in the false-color mapping among the trees. A legend for irradiance ranges from 0 to 500 Watts per meter squared. — Five planting designs (A-E) and their impact on plant architecture and light interception.

The authors found that while architectural plasticity under high planting density improved light interception via leaf area expansion, competition for light imposed by the design countered-balance this benefit in terms of carbon assimilation at stand scale.

Perez concludes, “the modelling tool presented in this paper paves the way to design plantation planting patterns based on in silico estimations of plant performance.”

READ THE ARTICLE:

Raphaël P A Perez, Rémi Vezy, Loïc Brancheriau, Frédéric Boudon, François Grand, Merlin Ramel, Doni Artanto Raharjo, Jean-Pierre Caliman, Jean Dauzat, When architectural plasticity fails to counter the light competition imposed by planting design: an in silico approach using a functional–structural model of oil palm, in silico Plants, Volume 4, Issue 1, 2022, diac009, https://doi.org/10.1093/insilicoplants/diac009

Processed data, as well as models code, are available from the authors upon request. Archimed-phi model is available online (https://archimed-platform.github.io/archimed-phys-user-doc/). An example configuration of simulation is available in Zenodo repository (doi:10.5281/zenodo.6246090).