Experimental setup
The soil was collected in November 2020 from the upper 30 cm of an agriculturally used Haplic Luvisol (Cai et al. 2016) located in Selhausen, Germany (50˚52′07.8′′ N, 6˚26′59.7″ E), with a silt loam texture (22% clay, 66% silt, 12% sand) (Weihermüller et al. 2007). After sampling, the soil was homogenized, sieved to < 2 mm, and air-dried. The columns (8 cm inner diameter and 45 cm high) were filled with air-dried soil and compacted to reach a dry bulk density of 1.4 g cm− 3, representative of field conditions. Importantly, bulk density was homogeneous across the soil profile as this could affect root development and infiltration of water.
The two experimental spring wheat lines (Triticum aestivum L.) examined in this study were UQR012 (SRS) and UQR015 (DRS), which were developed by backcrossing a donor source for narrow root angle to the high-yielding spring wheat cultivar Borlaug100, as reported by Rambla et al. (2022). The genotypes UQR012 and UQR015 were selected because they displayed contrasting seminal root angles (RA°: 110° and 66°) which made them ideal for investigating the effect of root distribution on the composition and function of microbial communities at different soil depths. The genotypes were planted in monocultures; a total of four seeds were sown per column and after emergence the two strongest seedlings were retained. A total of 13 columns per genotype were used, three of which were destructively sampled by extracting the soil out of the columns at 28, 35 and 42 days after sowing (DAS) and the remaining four 49 DAS, corresponding to the tillering, stem elongation, booting, and ear emergence stages, respectively. Plants were grown in a climate chamber under controlled conditions at light intensity of 1000 µM m− 2 s− 1 using a 12 h photoperiod (from 8 am to 8 pm). The temperature was set to 20°C during the day and 18°C at night, with humidity of 50%. The soil columns were saturated with deionized water by capillary action from the bottoms of the columns. The mean initial soil gravimetric water content (θsoil) was 0.40 cm3 cm− 3. During plant growth, water was added twice a week to keep the θsoil value between 0.30 cm3 cm− 3 and 0.44 cm3 cm− 3. On each sampling day, aboveground plant biomass was removed and the soil taken out of the columns and cut into horizontal segments of 9 cm thickness. After performing non-destructive analysis on selected segments (zymography), the rhizosphere was sampled by removing all roots with attached soil aggregates (< 5 mm) by hand (Gobran and Clegg 1996). Rhizosphere soil was separated from the roots by gently shaking for 1 min in a plastic container. The soil was stored at -20°C until analysis.
Soil zymography and image analysis
Soil zymography was performed on two segment surfaces representing soil depths of 0–9 cm (further referred to as topsoil) and 27 to 36 cm (further referred to as subsoil) according to Razavi et al. (2016). We did not perform zymography on the last segment, 36 to 45 cm, due to the impact of root accumulation at the bottom of the columns, especially at the later sampling times, and because of stagnant water conditions in this layer. The fluorogenic substrate 4-methylumbelliferone-β-D-glucopyranoside (Sigma Aldrich, Germany) was used to determine the spatial distribution of the BG activity. The substrate was dissolved in dimethyl sulfoxide (Sigma Aldrich, Germany) and diluted to a concentration of 5 mM using deionized and autoclaved water. A polyamide membrane filter with of pore size 0.2 µm (Sartorius, India) was soaked with the substrate solution and placed directly on the soil surface. Before placing the filter on the soil surface, the soil was slightly moistened to allow sufficient diffusion of the substrate into the soil (Gruber et al. 2021). The filter was incubated on the soil for 1 h in the dark at 20°C. Afterwards the filter was removed and exposed to UV light (365 nm) using the BioDOC Analyzer (Biometra, Germany). Pictures were taken (RICOH TV-200M 8–48 mm) with an exposure time of 75 ms and an enhancement of 5 using the BioDocAnalyze software. Small membrane snippets (3 x 1 cm) were soaked in solutions of increasing concentrations of 4-methylumbelliferone (MUF) (0, 0.05, 0.1, 0.15, 0.2 and 0.3 mM) for calibration. Enzyme activity was calculated by the volume of MUF solution, taken up by the polyamide membrane and its size (Razavi et al. 2016). For both, the zymograms and the calibration, the same filters and camera settings were used.
We used the open-access software ImageJ (NIH, USA) with the package Fiji (Version 2.1.0) for image processing. The zymograms were converted into 8-bit grayscale images and the background fluorescence of filters soaked with MUF-substrate solution were subtracted from the gray values of the images. The gray values of the images were transformed into amount of MUF per membrane area (pM mm− 2) using the calibration function (R2 = 0.99). A small piece of ordinary squared paper was used to set the scale for the observed area using the “Set Scale” function of ImageJ. To compare the effect of the different root architectures on BG activity, average enzyme activity was calculated using the histogram data of the zymogram after calibration and background correction.
Enzymatic hotspots were defined based on other studies as having gray values exceeding 50% of the mean gray value and equivalent to an enzyme activity of 11 pM mm− 2 h− 1 in this experiment (Heitkötter and Marschner 2018; Hao et al. 2022). Accordingly, enzymatic coldspots were defined as the areas that fell below 50% of the mean enzyme activity. The hot- and coldspot areas were calculated as percentages of the total soil surface area.
The BG activity gradient from the root center towards the surrounding soil was determined on selected roots with highest enzyme activity by using the “Plot Profile” function of ImageJ with a line width of five pixels (corresponding to 781 µm) on six to eight roots for each genotype, sampling time, and depth. The BG activity gradient was then plotted against the distance. The decrease in enzyme activity E(x) from the root center towards the surrounding soil was described using an exponential decay function:
$$E\left(x\right)={E}_{0}\cdot {exp}\left(-kx\right)+{E}_{bulk}$$
(1)
where E0 is the initial enzyme activity (pM mm− 2 h− 1) close to the root center, x is the distance to the root center (mm), k is a first-order rate coefficient (1/mm) of enzyme activity decrease and Ebulk represents the mean enzyme activity (pM mm− 2 h− 1) in bulk soil.
To determine the rhizosphere extent (Rext), the rhizosphere threshold Eᶲ based on the BG activity in the bulk soil was calculated using the following equation:
$${E}_{ᶲ}={E}_{bulk}+{2\sigma }_{bulk}$$
(2)
where σbulk is the standard error of the mean enzyme activity (pM mm− 2 h− 1) in bulk soil (Ebulk).
Ebulk and σbulk were determined by measuring the BG activity of non-root affected areas in the zymogram. Areas with an enzyme activity equal to or greater than Eᶲ were considered as rhizosphere soil. The extent of the rhizosphere in mm was calculated using the following equation:
$${R}_{ext}= -\text{l}\text{n}\left(\frac{{E}_{ᶲ}-{E}_{bulk}}{{E}_{0}}\right)/k$$
(3)
Potential enzyme activities in soil
The potential activity of three major enzymes involved in C-, N-, and P-cycling, i.e., BG (EC 3.2.1.21), NAG (EC 3.2.1.52) and AP (EC 3.1.3.2), were determined according to Marx et al. (2001) by using fluorogenic 4-MUF substrates. Substrates and MES buffer were prepared according to Poll et al. (2006). To determine the enzyme activities, 1 g of soil was dispersed in 50 ml of deionized and autoclaved water by an ultrasonic disaggregator (50 J s− 1) for 2 min. Afterwards, the suspensions were stirred on a magnetic stir plate and 50 µl aliquots were dispensed into 96-well microplates (PP F black 96 well, Greiner Bio-one GmbH, Germany). Subsequently, 50 µl of the autoclaved MES buffer and 100 µl of the substrate solution were added. For the standards, 50 µl of the soil suspension and an appropriate amount of standard solution and MES buffer were added to a final volume of 200 µl as reaction medium. The MUF concentrations of the standard solutions were 0, 100, 200, 500, 800, and 1200 pM per well. The plates were incubated under the exclusion of light at 30°C. After a pre-incubation period of 30 min the fluorescence intensity at 460 nm was determined by exposing the plates to a wavelength of 360 nm using a microplate fluorescence reader (Bio-Tek Instruments Inc., FLX 800, Germany). Fluorescence was measured at 0, 30, 60, 120 and 180 min after the 30 min pre-incubation period. Enzyme activity (in nM g− 1 h− 1) was determined using the linear correlation between the fluorescence intensity and the MUF concentrations of the standards.
Soil microbial carbon
Microbial carbon (Cmic) was determined using the chloroform fumigation extraction method by Vance et al. (1987). Two subsamples of 1 g soil were weighed from each sample. One subsample was fumigated in a desiccator with ethanol-free chloroform for 24 h to release Cmic. The remaining chloroform was then removed by flushing the desiccator 10 times with fresh air. Afterwards the fumigated and non-fumigated subsamples were extracted with 4 ml of 0.5 M K2SO4 solution, shaken at 200 U min− 1 for 30 min, and centrifuged at 4400 g for 30 min. After shaking and centrifugation, the supernatant was transferred into a scintillation vial using a 5 ml pipette equipped with a 20 µm filter on the tip to avoid inclusion of organic particles. The extracts were frozen at -20°C until measurement. Measurement of organic C in the supernatants was performed using the TOC-TNb Multi N/C 2100S Analyzer (Analytik Jena, Germany). Microbial biomass C was calculated by subtracting the C content of the non-fumigated from the fumigated extracts using the kEC factor of 0.45 according to Joergensen (1996) to correct for the extractable part of total C bound to the microbial biomass. In addition, the non-fumigated samples were used to calculate the extractable organic C (EOC).
Phospholipid and neutral fatty acids
The main microbial groups were determined by extracting phospholipid and neutral fatty acids (PLFA, NLFA) from microbial cell membranes. The lipids were extracted from the soil, fractionated, and quantified according to Bardgett et al. (1996) based on the method of Frostegard et al. (1991) and Bligh and Dyer (1959). In brief, 4 g of soil was mixed with Bligh & Dyer solution (ratio of chloroform: methanol: citrate buffer = 1:2:0.8) to extract the lipids. To separate the NLFAs from PLFAs a solid phase extraction with extraction columns (Bond Elut, Agilent Technologies, USA) was used. Afterwards, the PLFAs and NLFAs were transformed into fatty acid methyl esters (FAMEs) by alkaline methanolysis, as described by Kramer et al. (2013). For quantification, an internal standard of the FAME C24:1 (Sigma-Aldrich, St. Louis, MO, USA) was added to the samples prior to methanolysis. Following Kandeler et al. (2015), the fatty acids i15:0, a15:0, i16:0, 16:1ω7, i17:0, cy17:0, and cy19:0 were chosen to represent bacterial PLFAs. The sum of iso and anteiso fatty acids was used as an indicator for the gram-positive bacteria (PLFAGram+) and the sum of cyclic fatty acids were used as an indicator for gram-negative bacteria (PLFAGram−). The fatty acid 18:2ω6,9 was chosen to represent fungi (PLFAfun) (Federle 1986). Additionally, the NLFA 16:1ω5 was used as a biomarker for arbuscular mycorrhizal fungi (NLFAAMF) (Olsson et al. 1998). The extracted FAMEs were analyzed on an Agilent 8860 gas chromatograph equipped with a 5977B mass selective detector (MSD) (Agilent, USA), calibrated with a bacterial methyl ester mix (Sigma-Aldrich, USA) and individual standard FAMEs.
Statistical analyses
Statistical analyses were conducted in R (Version 4.2.0) (R Core Team 2020). Normality and homogeneity of the enzyme activity, enzymatic hotspot area, rhizosphere extent, abundance of the main microbial groups and microbial carbon were analyzed using the Shapiro-Wilk´s test and Levene test from the car package (Fox and Weisberg 2019). The significance of differences (α < 0.05) of enzyme activity, enzymatic hotspot area, abundance of main microbial groups, and the microbial carbon were tested by a linear mixed-effects model using the lme function from the nlme package, where the columns were considered as a random effect and genotype, depth, and time as fixed effects (Pinheiro and Bates 2000).
To model the enzyme gradient within the rhizosphere a nonlinear mixed-effects model was fitted to the BG activity with increasing distance to the root center. This nonlinear mixed-effects model was simplified based on the significance of factors and interactions to identify the most important components of the model and to reduce overfitting. Significance of difference was then tested by performing an ANOVA on the simplified model.
To examine the relative focus of microbial acquisition on C versus nutrients (N and P) and P versus N acquisition, the untransformed proportional enzyme activity vector lengths (Lv) and angles (θv) were determined using the following equations (Brandt et al. 2023; Moorhead et al. 2016):
$${L}_{v}=\sqrt{{{E}_{C-P}}^{2}+{{E}_{C-N}}^{2}}$$
(4)
$${\theta }_{v}=\text{a}\text{t}\text{a}\text{n}({E}_{C-P},{E}_{C-N})$$
(5)
In this context, EC−P indicates the relative activities of C- versus P-acquiring enzymes (BG/(BG + AP)), and EC−N represents the relative activities of C- versus N-acquiring enzymes (BG/(BG + NAG)). Using the arc tangent (atan), the vector angle of the line running from the origin of the diagram to the point (EC−P, EC−N) was calculated.