Fig 1.
Layout of the autonomous duckweed loading system presented in a), and schematic of the protocol sequence demonstrating the duckweed loading operation presented in b).
Fig 2.
Autonomous imaging system built in a growth chamber in isometric view a), and in top view b). The cameras take images from below the transparent stages holding plates. The grid on the transparent stages prevents plate misalignment. The linear actuator carries the imaging system to complete the imaging.
Fig 3.
Workflow of the phenotyping tool.
First, the raw image containing two well-plate images is split into individual well-plate images. Then, the wells are identified, and decomposed into individual images. Finally, color thresholding extracts duckweed features in a well including frond area, number, and greenness.
Fig 4.
Growth of L. minor in all 6,000 experimental units through time.
Frond area in all wells was measured once per day, and each line connects the area measurements of a single well across time. Lines for all 6,000 wells are plotted on top of each other, such that the high-density area (black, many lines plotted on top of each other) depicts the most common growth trend across all experimental units. Individual wells sometimes appear to grow from zero to positive frond area–these must have always contained some living tissue, as no new duckweeds were added. Wells could be scored as having zero living tissue if the living tissue did not meet the size or color thresholds in the processing software. For example, fronds infrequently do recover from initial yellowing, or, if they die, their internally held daughter fronds may remain alive and become visible in one or two days.
Fig 5.
Growth of duckweeds across nitrogen levels.
Each line is frond area in one well. Darker lines depict smoothed conditional means of plants inoculated with microbes (blue) and plants not inoculated with microbes (orange), averaged across other nutrient levels.