Résumé
We developed a 2U incubator and used it to grow the legume Medicago truncatula autonomously. This prototype was designed to become a payload in a 3U Cubesat nanosatellite; it therefore weighs only 1.2 kg and has a total consumption of less than 4 Watt. The incubator is equipped with many sensors to monitor its environment such as gas, humidity and heat sensors. It also contains actuators to modify the environment such as a flexible heater and a TiO2-based ethylene photocatalyst to remove plants ethylene production. The objectives are first to determine good growth conditions in this limited volume device and to allow better prototyping of a 2U incubator experiment. Second, to build a prediction model based on measurements of plants functional traits. Therefore, first we carried out a design of experiments to perform the germination and growth of M. truncatula at temperatures of 22 °C and 29 °C, under CO2 concentrations of 380 ppm and 10000 ppm and with two photoperiod regimes 16h/8h and 20h/4h. This gives a total of 8 experimental conditions tested in large climatic chambers during a 30-day period. The 2U experiment was performed at 26 °C, 380 ppm CO2 and with a 20h/4h photoperiod. This experiment lasted 62 days before being stopped to find that 2 large plants had grown for almost 52 days. Secondly, by using a principal components analysis, we observed that most of the plants functional traits variables explained the variability on the first principal component, with the exception of the surface area of the small leaves and the quantity of small leaves which explained the variability on the second principal component. The quantity of big leaves, which is an easily measurable variable with a camera, was very strongly correlated with the first principal component. These 3 variables and 2 others were included in the model equations used to predict the values of visible and hidden plants functional traits. This model has been successfully tested on the 2U incubator experiment allowing precise determination of fresh weight biomass under 20% error. The determination of the fresh weight of the roots, the fresh weight of the shoots, the amount of big and small leaves, the total amount of leaves, the surface area of big leaves and small leaves and the total surface area of the leaves was carried out.