Subjects
Twenty–four snow peas (Pisum sativum var. saccharatum ‘Carouby de Maussane’) were chosen as the study plants (see Table 1). P. sativum seeds were potted and kept at the conditions outlined below.
Table 1
No Support condition |
---|
N° | 8 |
Distance | – |
Age | 21.12d (± 1.5) |
N° of leaves | 5 (± 0.93) |
Support condition |
N° | 16 |
Distance | 12 cm |
Age | 23d (± 1) |
N° of leaves | 5 (± 1.67) |
Note. The age of the plant was expressed in days, and the N° of leaves refers to the mean, while the standard deviation is noted in parentheses.
Experimental conditions
P. sativum plants were tested in an environment in the presence (i.e., ‘Support’ condition) or in the absence of potential support (i.e., ‘No Support’ condition; Fig. 7A). The support was a 60 cm high wooden pole (i.e., the inground part was 7 cm long, while the aboveground part was 47 cm in height) positioned at 12 cm from the plant’s first unifoliate leaf (Fig. 7A).
Germination and growth conditions
Cylindrical pots (diameter 20 cm; height 20 cm) were filled with silica sand (type 16SS, dimension 0.8/1.2 mm, weight 1.4). The pots were watered and fertilized using a half-strength solution culture (Murashige and Skoog Basal Salt Micronutrient Solution; 10×, liquid, plant cell culture tested; SIGMA Life Science, Milan, Italy) and then watered with tap water as needed three times a week. Seeds were soaked in water for 24 hours and then placed in absorbent paper for 5 days to germinate. Once the seeds germinated, healthy seedlings of similar height were chosen and potted. Each pot was then enclosed in a growth chamber (Cultibox SG combi 80x80x160 cm; Fig. 7A) so that the seeds could germinate and grow in controlled environmental conditions. The chamber air temperature was set at 26°C; the extractor fan was equipped with a thermo-regulator (TT125; 125 mm–diameter; max 280 MC/H vents), and there was an input-ventilation fan (Blauberg Tubo 100–102m3/h). The two-fan combinations allowed for a steady air flow rate into the growth chamber with a mean air residence time of 60 seconds. The fan was placed so air movement did not affect the plants’ movements. Plants were grown with an 11.25-hour photoperiod (i.e., 5.45 am to 5 pm) under a cool white LED lamp (V–TAC innovative LED lighting, VT–911–100W, Des Moines, IA, USA or 100W Samsung UFO 145lm/W – LIFUD) that was positioned 50 cm above each seedling. Photosynthetic Photon Flux Density at 50 cm under the lamp in correspondence of the seedling was 350 umolph/(m2s) (quantum sensor LI–190R, Lincoln, Nebraska USA). Reflective Mylar® film of chamber walls allowed for better uniformity in light distribution. The experimental methodology was applied to the single plants grown individually in a growing chamber.
Video recording and data analysis
For each growth chamber, a pair of RGB–infrared cameras (i.e., IP 2.1 Mpx outdoor varifocal IR 1080P) were placed 110 cm above the ground, spaced at 45 cm to record stereo images of the plant. The cameras were connected via Ethernet cables to a 10–port wireless router (i.e., D–link Dsr–250n) connected via Wi–Fi to a PC, and the frame acquisition and saving process were controlled by Cam Recorder software (Ab. Acus s.r.l., Milan, Italy). To maximize the contrast between the anatomical landmarks of the P. sativum plants (e.g., the tendrils) and the background, black felt velvet was fixed on some sectors of the walls of the boxes, and the wooden supports were darkened with charcoal. Each camera’s intrinsic, extrinsic, and lens distortion parameters were estimated using a Matlab Camera Calibrator App. Depth extraction from the single images was carried out by taking 20 pictures of a chessboard (squares with 18 mm of side, 10 columns, 7 rows) from multiple angles and distances in natural non–direct light conditions. The same chessboard used for the single camera calibration process was placed in the middle of the growth chamber for stereo calibration. The two cameras then took the photos to extract the stereo calibration parameters. In accordance with the experimental protocol, a frame was synchronously acquired every 3 minutes (frequency 0.0056 Hz) by the cameras. An ad hoc software (Ab. Acus s.r.l., Milan, Italy) developed by Matlab was used to position the markers and track their position frame–by–frame on the images acquired by the two cameras to reconstruct the 3D trajectory of each marker41. All leaves developed by the plants were analyzed from the germination of the seed until the plant collapsed (i.e., ‘No Support’ condition; Fig. 1C) or coiled the support (i.e., ‘Support’ condition; Fig. 1B). For all leaves in both experimental conditions, the initial frame was defined as the frame in which the tendrils started to develop, and they were visible from the apex. The end of movement for the uncoiled leaves was defined as the frame in which tendril(s) stopped producing its own circumnutation. For the last leaf (i.e., Llast; Fig. 1B, C) developed by the plants, the end of movement was defined as the frame before the plant collapsed in the ‘No Support’ condition or in the ‘Support’ condition when the tendrils started to wrap around the support. The markers on the anatomical landmarks of interest of the plants, namely the apex, the junction of the tendrils, and the tendrils, were inserted post–hoc (Fig. 7B). The markers were also positioned on the support (i.e., on both the lowest and the highest points of the support), the origin of the plant, and internodes as reference points (Fig. 7B). The tracking procedures were first performed automatically throughout the time course of the movement sequence using the Kanade–Lucas–Tomasi (KLT) algorithm on the frames acquired by each camera after distortion removal41. The tracking was manually verified by the experimenter, who checked the position of the frame of the marker frame by frame. The 3D trajectory41 of each tracked marker was computed by triangulating the 2D trajectories obtained from the two cameras (Fig. 7C).
Dependent measures
The dependent variables specifically tailored to test our topic based on previous evidence12–17,19,20,41 was:
-
Spatial trajectories: this measure allows us to describe circumnutation in both qualitative and quantitative terms.
-
Movement time (min): The interval between the movement’s beginning and end. That is when the plant encountered the potential support (i.e., ‘Support’ condition) or collapsed (i.e., ‘No Support’ condition).
-
The average circumnutation velocity (mm/min).
-
The distance from the origin of the plant to the center of circumnutation: Euclidean distance (mm) between the Circumnutation Center (i.e., the geometric center of gravity in the X–Z plane computed as the mean of each coordinate for all the points constituting the circumnutation) and the plant origin in the X–Z plane.
-
Area of the circumnutation (mm2) as the sum of pixels with a value equal to 1, obtained from the binarization of the circumnutation trajectory.
-
Average circumnutation acceleration (mm/min2).
-
Number of circumnutations performed by the plant during the entire movement time.
-
Number of switch directions during each circumnutations (clockwise, counterclockwise, and None). For each circumnutation, the sum of all the angles between the movement vector at time t and the movement vector at time t + 1 is calculated. The direction, then, is determined according to the following logic: if the resulting sum is equal to 2π ± 1.2, then the direction is counterclockwise, or else if the resulting sum is equal to − 2π ± 1.2, then the direction is clockwise. For all other cases, no direction is assigned.
The kinematical indices were scaled for standardization purposes.
Statistical analysis
Data analyses were computed in the R environment42. Categorical variables, such as Experimental Phase (‘PRE,’ ‘POST’) or Leaf (‘Third last,’ ‘Second last,’ ‘Last’), were created for analysis. For the investigation of Q1 (See Results section), four mixed effects linear models (i.e., lmer; ‘lme4’)43 were fitted for each kinematical variable, setting the interaction between Experimental Condition (‘Support,’ ‘No Support’) and Experimental Phase, as well as random intercept (plants, 24 specimens) and random slope (Experimental Condition). The apex of each plant retrieved data, and the total number of observations considered for each model was equal to 3117. Dependent variables were scaled for standardization purposes. For answering Q2, four Generalized Additive Models (i.e., gam) from the ‘mgcv’ R package44 were fitted (with the REML method) for the investigation of non–linear relationship between each of the four kinematical variables and the other indices, controlling for Experimental Condition and Experimental Phase. Predictors were smoothed to study the complexity of the non–linear relationship with the dependent variables, which was scaled as well as the predictors. The smoothed random intercept of each plant was considered (n = 24). Data were retrieved by the apex of each plant, and the total number of observations considered for each model was equal to 3115. The supplementary materials represent relationships among kinematical indices by fitting B–spline curves with 3 degrees of freedom. For the investigation of Q3, four mixed effects linear models (‘lme4’)43 were fitted for each kinematical variable, testing the interaction between Experimental Condition, Leaf, and the anatomical landmark of the plant (‘Apex,’ ‘Tendrils’), also setting random intercept (plants, 24 specimens) and random slope (Experimental Condition). Data were retrieved by the apex and tendrils (as a single object) of each plant, and the total number of observations considered for each model was equal to 5895. Again, dependent variables were scaled. To investigate Q4, we fitted a mixed effects linear model to test the interaction between Experimental Condition and Leaf on variations in the total number of circumnutations, while also setting plants as random intercepts. Data were retrieved by the apex and tendrils (as a single object) of each plant, and the total number of observations considered for each model was equal to 478. Contextually, a generalized mixed effect linear model (i.e., glmer) was fitted to the data to investigate the interaction between Experimental Condition, Leaf, and Switch Direction (i.e., ‘Clockwise,’ ‘Counterclockwise’) on variations in the total number of switches, setting a Poisson family distribution and plants as random intercepts. The present analysis did not consider switches with undetermined directions. Finally, movement time was investigated by fitting a mixed effect linear model, setting the ‘Movement time’ of the leaves dependent on the interaction between the leaf and experimental condition. The apex of each plant retrieved data, and the total number of observations considered for each model was equal to 3117. Considered the main interest in the interaction effects, Type 3 Sum of Squares was considered for deriving statistical results from lmer and glmer models. (R package ‘car’)45. Post hoc analyses were computed through the pairwise contrast test of the ‘emmeans’ R package46 when needed. For descriptive purposes, relationships among each kinematical index and Experimental Time are represented by fitting B–spline curves with 3 degrees of freedom. Descriptive graphics and model plots were developed via ‘ggplot2’ R package47 and are retrievable in the Supplementary Material.