Application of high throughput phenotyping via PlantScreen System to study early plant stress responses to progressive drought stress Klára Šimková, Sharmila Madhavan, Zuzana Benedikty , Diana Santelia and Martin Trtílek Photon Systems Instruments, Drásov, Czech Republic. Introduction Discussion Drought and salinity in soils are currently one of major problems in agriculture with rapidly growing importance in last few years. The use of advanced phenotyping that offers application of automated, high-throughput methods for characterisation of plant architecture and performance, has the potential for improvement of breeding process of different important crops. Conveyor and robotic PlantScreen System developed in PSI is designed for high throughput plant phenotyping in various species from Arabidopsis to crop plants. Recently we have used the PlantScreen System to study response of various Arabidopsis starch related mutants to drought stress conditions. We have optimized the screening conditions and the image processing analysis to obtain quantitative assessment of plant complex traits such as growth, development and physiological status. By using combination of chlorophyll fluorescence imaging, thermal imaging and morphology image analysis we have recognized a set of parameters that could serve as early stress markers for characterization of plant performance under adverse environmental conditions such as drought. Analysis of plant performance under water limiting conditions and better understanding of plant responses under drought conditions has in past years become major a major focus of research and investments. Primary ultimate goal of sustainable farming would be developing crops with improved water use efficiencies. Here we present rapid high-throughput non-invasive approach to assess numerous phenotypic traits of soil-grown plants during progressive drought stress treatment. High-throughput phenotyping platform, the PlantScreen System, was used for automated weighting,RGB morphometric analysis, kinetic chlorophyll fluorescence assesment and leaf temperature determination via thermal imaging. In addition hyperspectral analysis was used to determine water content in water-limited plants. Combinatorial statistical analysis was used to discriminate early drought stress marker. Here we show that the changes in light-adapted photosynthetic parameters during increasing water deficit monitored via chlorophyll fluorescence imaging can be used as primary marker for the rapid assessment and comparison of the relative viability of Arabidopsis plants during progressive drought. Methods RGB and structural imaging PlantScreen System was used for the high-throughput automated phenotypic analysis. Plants were grown in 12h-12h light conditions under cool-white LED illumination of 150 µE in standardized amount of given soil. When plants reached 6th leave stage last watering was performed. 3 days later the phenotypic analysis in high-throughut PlantScreen System was initiated. For next 10-11 days RGB structural imaging, chlorophyll fluorescence and thermal imagigng were monitored twice a day. Here we present data from one out of 4 biological experiments performed. Original image Barrel distortion correction Mask detection Chlorophyll fluorescence kinetic imaging Area Perimeter Roundness Compactness Leaf development tracking Greening index RLGR PAM light LED panel for kinetic imaging Fluorescence kinetics measurement FO, FM, FV, FO', FM', FV', Ft Max quantum efficinecy Fv/Fm Photochemical quenching Non-photochemical quenching Vitality index FV'/FM', PhiPSII , qN, qP PlantScreen Conveyer System Max light adapted fluorescence (Fp) Acclimation chamber/ weighing-watering station IR imaging station/ Hyperspectral imaging station Time (us) Time (us) Background subtraction Time (us) Time (us) Time (us) Time (us) Time (us) Time (us) Time (us) Time (us) Time (us) Time (us) Mask application and bckgr substraction Controlled cultivation environment Color segmentation Time (us) Time (us) Time (us) Time (us) Time (us) Time (us) Time (us) Time (us) Fluorescence quenching kinetics RGB imaging/Chlorophyll fluorescence imaging station Weighing and watering Hyperspectral imaging Operation software/PlantScreen database Automated weighing Original image Plant scale reflectance profile Weight of the pot Absolute water content Nutrient delivery Ecotoxicity testing 3D reconstruction Thermal imaging Water content NDVI PRI ..... Original thermal image 3D model reconstruction Reflectance histogram Binary image from RGB camera Watering in relative and absolute amount Temp (°C/K) Leaf Count Area Perimeter Orientation Plant Width Plant Height Bounding box Specific reflective indices Leaf temperature Leaf temperature change kinetics IR edge Days 3D model single leaf analysis Pattern recognition for mask matching IR image with overlaid RGB mask Results Morphometric analysis Days upon stress initiation Chlorophyll fluorescence imaging 12 1000 D1 800 1000 Series1 800 600 500 600 400 500 Control plants 27 28 29 31 32 33 Days upon germination 4 3 2 1 0 11 Area Normalised a.u. 5 3 4 RFD 2 900 1 800 9 10 8 1 700 0 0 2 4 6 8 10 12 Days upon stress initiation 600 11 9 3 4 Control Drought 3 30010000000 15000000 20000000 0 50000000 10000000 Time (us) Time (us) D5 RFD 700 0 DryF4 6 Control 4 Drought 100 9 4 0 8 10 6 Fig.5 Rapid rehydration stress was used to asses rate of water content loss by water absorption measuremnt with SNIR hyperspectral camera. Blue color refers to water content as detected by absorbance at 1440 nm and 1920 nm. 1 2 DryF3 RFD 400 DryF2 0 DryF3 DryF2 300 Time (us) 7 9 4 8 0 Fig.2 Among all parametres analysed area, compactness, perimetr and SOL 0 50000000 10000000 15000000 20000000 wereselected as useful early discriminants between control Time and (us) stressed group. Acknowledgements This work was carried out at Photon System Instruments (Czech Republic), with the financial support through HARVEST , FP7 Marie Curie Initial Training Network. 1 1 0h 0.9 300 200 2 20000000 D8 DryF4 1000 6 900 4 800 2 DryF3 15000000 100 Fig.4 Fp derived parametres can be used as primary discriminants between stressed 0 and control group already 7-8 days upon stress50000000 initiation. 10000000 15000000 20000000 0 0.5 7 50000000 0.5h 0.9 0.8 0.8 0.7 0.7 0.7 0.6 0.6 0.6 0.5 0.4 0.5 0.4 0.5 0.4 0.3 0.3 0.3 0.2 0.2 0.2 0.1 0.1 0.1 1000 1200 1400 1600 1800 2000 wavelength [nm] 2200 2400 0 800 2600 1 1000 1200 1400 1600 1800 2000 wavelength [nm] 2200 2400 0 800 2600 1 2h 0.9 4h 0.9 0.8 0.7 0.7 0.7 0.6 0.6 0.6 0.5 0.4 0.3 0.2 0.2 0.2 0.1 0.1 0.1 1000 1200 1400 1600 1800 2000 wavelength [nm] 2200 2400 2600 1600 1800 2000 wavelength [nm] 2200 2400 2600 1000 1200 1400 1600 1800 2000 wavelength [nm] 2200 2400 2600 5h 0.4 0.3 0 800 1400 0.5 0.3 0 800 1200 0.9 0.8 0.4 1000 1 0.8 0.5 1h 0.9 0.8 0 800 200 1 11 0 400 10 8 1.5 0 500 DryF2 SOL 0.5 500 0 7 Perimeter 0 600 4 1000 20000000 2 0 7 15000000 4 1 10 3 10000000 Time (us) DryF4 6 4 1 100 400 D1 3 2 11 200 Fig.3 Time course of chlorophyll fluorescence transients signatures in control and 200 stressed plants measured by quenching protocol .After 7-8 days maximum 100 fluorescence in light adapted state 0(Fp) differed between stressed and control 0 50000000 10000000 15000000 20000000 group. Time (us) 50000000 Compactness 3 0.5 300 0 Fig.1 Time course of plant development and stress response. Binary image is further used for morphometric analysis. Area 0 400 500 100 0 25 500 200 0 Drought Control D8 600 600 300 100 Control D8 700 700 400 200 23 800 Control D8 1000 Series1 900 Series2 800 700 300 22 900 900 Series2 700 D5 reflectance 11 reflectance 10 reflectance 8 reflectance 7 reflectance 6 Reflectance 4 reflectance 2 Stressed plants 1 Hyperspectral analysis 2200 2400 2600 0 800 1000 1200 1400 1600 1800 2000 wavelength [nm] Wavelenght (nm) Fig.6 Rapid rehydration stress was used to asses rate of water content loss by water absorption measuremnt with SNIR hyperspectral camera.