To assess the spatiotemporal variability of respiration along with root system development of Lupinus albus and changes in soil water content, we conducted three experimental time series. First, we measured rhizosphere pCO2 daily with soil water content kept constant and statistically located hotspots of respiration activity during 16 days. Second, we quantified the diurnal variation of respiratory activity by repeated measurements of pCO2 during the photoperiod of selected days within this period. Finally, we conducted a 3-days drying-rewetting experiment to investigate the sensitivity of CO2 and O2 concentration to fast changes in soil water content. After these 19 days, we harvested and analyzed the roots from regions where pCO2 strongly increased or pO2 strongly decreased after rewetting to correlate root tissue N content and respiratory activity.
Rhizotron preparation and plant growth
We prepared five glass rhizotrons (150 mm x 150 mm x 15 mm) with planar optodes sensitive to CO2 and two of them additionally with O2-sensitive optodes. The CO2 optodes (range: 1-25 % pCO2, size: 80 mm x 104 mm, product code: SF-CD1R, PreSens GmbH, Regensburg, Germany) were equilibrated in buffer solution (pH = 7.5) over night and then glued to the inner front windows (plants L1 – L5). The O2 optodes (size: 130 mm x 105 mm, manufactured as described in Rudolph et al., 2012) were attached to the inner back sides of two of the five rhizotrons (plants L4 and L5). Sandy soil (91 % sand, 8 % silt and 1 % clay, calcium acetate lactate (CAL) extractable P 8.6 mg kg-1, total N 0.01 %, total C 0.13 %, and pH(CaCl2) 7.6) was sieved to < 2 mm and filled horizontally into the rhizotrons (mean bulk density: 1.45 g cm-3). Seeds of white lupine (Lupinus albus) were sterilized in 70 % ethanol and planted after germination. Each plant was initially watered with 85 ml of a nutrient solution (containing 7% N, 3% P2O5, 6% K2O and micronutrients, as described in Rudolph-Mohr et al., 2017). Water was added to obtain an initial volumetric water content of 0.30 cm³ cm-³, which is equivalent to 77 % of saturation water content. After plant emergence, a gravel layer of 10 mm was placed at the soil surface to minimize evaporation. The water content was re-adjusted every morning to 0.30 cm³ cm-³ by irrigating from the top; no further nutrients were supplied throughout this experiment to obtain P-deficient conditions and stimulate cluster root development. The lupines were grown under controlled conditions in a plant growth chamber (temperatureday 24 °C, temperaturenight 19 °C, 14 hours photoperiod with a light intensity of 300 µmol m-2 s-1, relative humidity 60 %). Light intensity was increased from 0 % to 100 % between 6 a.m. and 10 a.m. and ramped down again to 0 % between 4 p.m. and 8 p.m.; temperature was changed between 19 °C (night) and 24 °C (day) accordingly. All samples were kept in an upright position, so roots distributed in soil towards both sides of the rhizotrons. The rhizotrons were covered with aluminum foil to protect the optodes from photobleaching.
Imaging of CO2 and O2 concentration
CO2 concentration was monitored with VisiSensTD, a commercial 2D fluorescence imaging and readout system (PreSens Precision Sensing GmbH, Regensburg, Germany). The CO2 optodes contain two fluorescent dyes (one sensitive to changes in pCO2, the other acting as a reference dye). A ring light source (built into the camera lens) and two external blue LEDs (wavelength 450 - 550 nm) were used to excite the fluorescent dyes. The fluorescence intensity was captured with an RGB camera (1292 x 964 pixels) at an exposure time of 70 ms and the signal ratio of the red and green channel (red:green ratio) was stored pixelwise. To convert this information into CO2 concentration (pCO2 in %), a calibration curve was fitted. For calibration, two pieces of CO2 optode were equilibrated overnight in a buffer solution (pH = 7.5, ionic strength = 40mM) and then fixed inside a small glass box filled with a similar buffer solution. The solution was flushed with gas mixtures of stepwise increasing CO2 concentration between 0 % and 25 % pCO2 and images were captured every 60 seconds at each calibration point until the signal was stable (taking between 15 and 20 minutes per concentration step). The calibration curve was fitted using the software VisiSens AnalytiCal (PreSens Precision Sensing GmbH). Fluorescence images (pixelsize 213 µm) were captured and directly converted to pCO2 maps in the software VisiSens AnalytiCal via the calibration curve.
The oxygen optodes were prepared according to Rudolph et al. (2012), with platinum (II) 5,10,20,-terakis(2,3,4,5,6-pentafluorophenyl)porphyrin as fluorescent dye incorporated into a polystyrene matrix. The optodes were calibrated in water with O2-concentration between 0 mg L-1 and 10 mg L-1 and a calibration curve was fitted based on the measured fluorescence intensities (as described in Rudolph et al., 2012). Fluorescence signals after excitation with UV light (type 215 L, Peqlab, Erlangen, Germany) were captured with a camera (Kappa DX 4C-285 FW) with a 500 nm long-pass filter and a cooled CCD sensor (1392 x 1040 pixels). The gray-value images (pixel size: 219 µm) were converted to O2-concentration maps based on the fitted calibration curve in MATLAB R2020(a) (The MathWorks).
Time series of rhizosphere pCO2 at constant water content and diurnal variation of rhizosphere pCO2 (experiment 1 & 2)
Experiment 1.We monitored pCO2 in the soil every day until day 16 after planting (DAP 16) to be able to identify hotspots of respiration amongst the growing root systems and the rhizosphere. As the only study to date applying CO2 optodes in unsaturated soil (Holz et al. 2020) suggests that measured magnitude of rhizosphere CO2 concentration is strongly influenced by soil moisture, we kept the volumetric soil water content constant at 0.30 cm³ cm-³ by irrigation every morning at 8:30 a.m. and always conducted the measurements 30 minutes after adjusting the water content to enable comparisons across plant individuals and root replicates.
Experiment 2. Additionally, we explored diurnal variations of respiration by measuring rhizosphere pCO2 in the morning (9:00 a.m.), at noon time (1:00 p.m.) and in the late afternoon (5:00 p.m.) on DAP 5, 8, 12 and 14 of experiment 1. On these four days watering to 0.30 cm³ cm-³ took place at 8:30 a.m. as usual, but was re-adjusted also 30 minutes prior to the second and third measurement of the day, if water content varied by more than 0.02 cm³ cm-³. The first measurement in the morning took place 3 h after start of illumination, the second 3 h after reaching 100% illumination and the last measurement 3 h before the light in the plant growth chamber was turned off for the night.
Changes of pCO2 and pO2 after rewetting of dry soil (experiment 3)
Experiment 3. The third section of the experimental time series aimed for quantification of pCO2 and pO2 in the rhizosphere following a fast increase in soil moisture. For that we stopped irrigation on DAP 16 and in the following conducted a drying-rewetting experiment. Water content declined to 0.10 cm³ cm-³ (26 % of saturation water content) on DAP 19 and we measured CO2 concentration (all five plants L1-L5) and O2 concentration (plants L4 and L5) in the dry soil. Afterwards, the rhizotrons were rewetted to 0.30 cm³ cm-³ (77 % of saturation water content) from the bottom. Then CO2 and O2 concentrations were measured directly (0.2 h) after rewetting as well as 1 h, 2 h, 3 h, 4 h and 5 h after increasing soil moisture.
To limit stress during our experiments, plants were only briefly taken out of the plant growth chamber to a darkroom for imaging and returned directly afterwards. In the darkroom the rhizotrons were placed in a sample holder mounted on a table to ensure that they were always aligned in the same position relative to the camera.
Measurement of root position and cluster root development
Since the CO2 optodes include an optical isolation layer, it was not possible to capture optical images of the precise location of roots systematically without removing the optode. To avoid disturbing gas transport dynamics in the soil, we did not open the rhizotrons or remove the optodes until the end of all experiments. However, several cluster and lateral roots or root segments were visible through the optode and we could trace their position with a pen on the glass window. These regions were later used for quantitative analysis of root zone pCO2 in experiment 1 and 2. After the drying-rewetting experiment on DAP 19, we opened the rhizotrons, removed the CO2 optodes and captured images of the exposed root systems (plant age: 21 days) to locate the position of all roots growing along the optodes. Plants L1, L2, L3 and L5 had developed several cluster roots close to the CO2 optode. Just plant L4 grew only lateral roots without clusters close to the CO2 optode (Fig. S1). The O2 optodes are semi-transparent and therefore we could trace roots directly from images taken at ambient light conditions. Both L4 and L5 grew cluster roots close to the oxygen optode. After imaging the opened rhizotrons, we washed the root systems carefully to remove soil particles and captured images to determine the extent of cluster root development amongst the entire root system.
Root sampling for nitrogen (N) content analysis
Based on the fluorescence image time series captured during the drying-rewetting experiment (DAP 19), we selected regions of considerably higher and lower respiratory activity (considering both CO2 and O2 concentration) and took root samples there. We did not distinguish between cluster and lateral roots during sampling, but only selected roots growing close to the optodes. The sampled roots and root segments were dried at 60 °C for at least 48 h and then ground for analysis. Root carbon (C) and nitrogen (N) contents as well as the C:N ratio were determined in two replicates per region by elemental analysis (Euro EA 3000 Elemental Analyser, HEKAtech GmbH, Wegberg, Germany).
All images were registered with the Plugin “Stackreg” in ImageJ prior to further analysis. CO2 concentration (in % pCO2) was directly calculated from the fluorescence images in the VisiSens AnalytiCal software and saved as TIFF images. O2 concentration was calculated in MATLAB R2020a as described in Rudolph-Mohr et al. (2017) and converted to % pO2.
We statistically located hotspots of CO2 concentration in the rhizosphere following the approach suggested by Bilyera et al. (2020). First, we converted the CO2 image time series of each plant to 8-bit gray value maps of pCO2 and saved the histogram of gray values of each image (MATLAB R2020a). The gray value distribution was then statistically split into two distributions (package “mixtools” in RStudio, Bengalia et al., 2009) to separate hotspots from background. Pixels were classified as hotspots when the gray value was higher than the mean + 3SD (three times the standard deviation) of the background pixel values (Bilyera et al. 2020). The hotspot area (in mm²) was calculated by multiplying the number of hotspot pixels by the pixel size and was compared to the total area covered by the optode.
Diurnal variation of pCO2 was compared in selected regions of interest (10 x 10 pixel, approx. 4 mm²) close to roots that were visible through the optode (cluster roots: n = 7, lateral roots: n = 18 on DAP 14) and within the bulk soil (n = 25). To compare rhizosphere respiration during the drying-rewetting experiment on DAP 19, we first segmented roots growing close to the optodes from the images of the exposed root systems captured after opening the rhizotrons (“SmartRoot” Plugin in ImageJ, Lobet et al., 2011). We then interactively selected a total of 47 non-overlapping roots or root segments from the binary images obtained from segmentation (“drawpolygon” and “poly2mask” function, Image Processing Toolbox, MATLAB R2020a). CO2 and O2 concentration as a function of distance to the root surface was calculated using the Euclidean distance transform (via “bwdist” function in Matlab). We graphically estimated the extent of CO2 accumulation resp. O2 depletion zones at different volumetric soil water contents by fitting local regression curves (function “loess” in RStudio) to the mean CO2 resp. O2 concentration with increasing distance from the roots.
Measured CO2 and O2 concentration in the root zone and the bulk soil were analyzed for normality and homogeneity of variances applying Shapiro Wilk’s test and Levene’s test, respectively. Differences between cluster and lateral roots as well as the effect of soil water content were tested for statistical significance using Kruskal-Wallis test followed by a Wilcoxon test. C and N content and C:N ratio of roots from regions of high vs. low respiration was compared pairwise also applying a Wilcoxon test. All statistical tests were computed at a significance level of α < 0.05 in RStudio (R Core Team, 2020).