Experimental design
The experiment was performed using 60 columns (sewer base pipes) with an inner diameter of 18.9 cm and a height of 80.5 cm. The first 60 cm of the columns were filled with subsoil while the upper 20.5 cm were filled with topsoil. The bottom of the columns was covered with a perforated lid. The soil used for the experiment was an Albic Luvisol (according to the World Reference Base for Soil Resources) and was collected from the Teaching and Research Station of the Albrecht Daniel Thaer-Institute of Agricultural and Horticultural Sciences of Humboldt-University of Berlin (location coordinates: 52°28′07.5″N 13°17′43.5″E) with help of a mini excavator. The subsoil was collected in a depth range of 40–75 cm, and the topsoil was collected in a depth range of 0–25 cm. Soil chemical and physical properties are given in Table 1. Prior preparation of the columns, both the subsoil and the topsoil were dried, and the topsoil was sieved with a sieve mesh size of 0.5 cm and the subsoil was sieved with a sieve mesh size of 0.3 cm.
Table 1 Soil physical and chemical properties of the topsoil and subsoil. Potassium, Mg, and P are given in mg 100 g−1 soil. Total carbon (Ct) and total nitrogen (Nt) are given in % In a first step, the lowest part of the column was filled with a gravel-sand mixture of 10 cm height (Fig. 1). Afterwards, subsoil was filled in five steps resulting in 5 subsoil layers with a soil depth of 15–24 cm, 24–34 cm, 34–44 cm, 44–54 cm, and 54–64 cm (Fig. 1). In order to simulate environmental conditions with minimized water supply in the subsoil part of the columns, a subsoil moisture of 10% of the maximum water holding capacity (WHC) was set in 30 columns. This experimental group was referred to as WHC10%. In the other 30 columns, a subsoil moisture of 30% of the maximum WHC was set, referred to as WHC30%. For this, prior filling the soil into the columns, the WHC100% of the sub- and the topsoil were determined by assessing the soil weight using calibrated cylinders of the fully water saturated soil (by water-saturation for 3 h with a subsequent draining of 2 h) and of the dry soil (by drying for 24 h at 105° C in a drying chamber). The WCH100% was then calculated with the formula:
$${\mathrm{WHC}}_{100\mathrm{\%}} = \frac{\mathrm{Weight saturated soil}-\mathrm{Weight dry soil}}{\mathrm{Weight dry soil}}* 100$$
With help of this, the required water amount was calculated for setting a soil moisture of WHC10% or WHC30%, respectively.
When preparing the columns, the subsoil was compacted to a density of 1.7 g cm−3 while the topsoil was compacted to a density of 1.4 g cm−3. This was realized with help of a workshop press.
After filling the subsoil into the columns, artificial vertical pores with a diameter of 5.3 mm were created with help of an iron rod in the subsoil part. Previous studies have shown that this diameter is large enough to allow incubation with adult individuals of L. terrestris (Dresemann et al. 2018). Five different pore densities were set in each soil moisture regime: 0, 1, 2, 3, and 4 pores column−1 (Fig. 1). The pores were drilled in symmetrical distances to the center of the columns and the wall of the columns. The present experiment finally had 10 different treatments with 6 replications, resulting in 60 columns (Fig. 1). Next, one earthworm (Lumbricus terrestris L.; obtained from proinsects, Minden, Germany) was introduced into each pore. The subsoil and the pores were covered with a freshly harvested grass-clover mixture from the Teaching and Research Station of the Albrecht Daniel Thaer-Institute of Agricultural and Horticultural Sciences of Humboldt-University of Berlin. The purpose of introducing the earthworms into the pores was to line the pores with earthworm excrements in order to make the pore properties more natural for the subsequent plant growth. After 28 days, the earthworms were casted out of the columns by creating a heating gradient with help of a self-constructed heating chamber (ESM_1.pdf). A temperature of approximately 41 °C was reached in the lower parts of the columns, encouraging the earthworms to leave the columns upwards. The heating chamber was placed in a daylight-free environment as earthworms are very light-sensitive. Only a red lamp was used as light source in order to check if an earthworm left a pore. Each heating procedure lasted four hours. In order to treat all columns equally, also those columns without pores were heated for the same time. Soil water was lost by the heating procedure. Thus, the columns were weighed before and after the heating and water equal to the weight difference was added on the top of the subsoil part that seeped into the soil after a while.
After casting out the earthworms, the upper column part was fixed on top of the lower part with duct tape, and the subsoil layers were covered with a 1 cm thick coarse gravel layer in order to prevent the topsoil from trickling into the pores and to prevent capillary rise from the sub- to the topsoil. Finally, the topsoil was added in two steps on top of the coarse gravel layer with a final height of 14 cm. The soil moisture in the topsoil was set to WHC30% in all 60 columns.
In order to document the soil moisture course, four time domain reflectometry (TDR) probes were installed in the subsoil part of one column of each treatment. The first probe was installed 14 cm above the column bottom. The further three probes were placed in a distance of 10 cm from the lower probe. The soil moisture was recorded three times per week from 18th of April until 27th of May 2019. The respective soil moisture values can be found in the supplementary file ESM_2.xlsx.
Finally, 12 plants per column of spring wheat (Triticum aestivum L. “Chamsin,” KWS, Germany) were sown on 18th of April 2019. All 60 columns were randomly arranged and plants grew for 60 days in a covered but translucent outdoor installation. Mean temperature course, the relative humidity, and the sunshine duration during this growth period are given in ESM_3.pdf. After 30 days, all columns were completely randomized again. In the first 40 days of growth, all plants were watered via the topsoil equally according to demand. The soil-moisture in the subsoil remained unaffected by this (ESM_2.xlxx). During the last three weeks of plant growth, watering was stopped in order to simulate lacking precipitation. Holes were drilled into the bottom of each column wall and the columns were placed in shells (23 cm diameter and 7 cm high) which were filled with water to a height of 5 cm in order to simulate a water source in deeper soil layers. The gravel sand mixture in the lowest part of the columns prevented capillary water rise.
Due to a mildew (Blumeria graminis) infestation, a plant protection agent treatment with the agent Adexar® (active ingredient: Xemium; BASF, Ludwigshafen, Germany) took place on 15 May 2019.
Shoot and root sampling
On 17th of June 2019 (60 days after sowing), all shoots were cut at the base. At this time point, most of the plants had reached the end of ear emergence. The fresh weight of all shoot biomasses column−1 was recorded, and subsequently the plants were cut into small pieces. Each sample was homogenized, a subsample was taken and dried at 60 °C in a drying chamber for four days for later mineral analyses. The rest of the sample was dried at 105 °C in a drying chamber for four days in order to determine the dry substance and the dry matter yield.
From 18th until 25th of June 2019, sampling of the soil layers started. First, the upper column section was separated from the lower section and the topsoil was removed. Adhering gravel from the underlying gravel layer was removed, and each topsoil layer was packed in plastic bags and stored in a cooling chamber at 6 °C until the roots were washed out. Afterwards, the lowest gravel-sand layer was removed. Then, the five subsoil parts, starting from the lower part, were pressed out in steps of 10 or 9 cm, respectively, with help of the workshop press. Similar to the topsoil, each subsoil layer was packed into a plastic bag and subsequently stored in a cooling chamber at 6 °C until further processing.
Subsequently, the roots were washed out of the top- and subsoil layers using a sieve with a mesh size of 0.63 mm. The coarser roots were removed from the sieve using tweezers. The residue was re-suspended in water in a petri dish and the finer roots were separated with tweezers. The roots obtained in this way were placed in a plastic container filled with distilled water and stored in the refrigerator at 8 °C until root length scanning for a maximum of two days.
For the root scanning, the washed roots were evenly distributed in a plastic frame and fully covered with distilled water. The scanning was performed with a flat-bed scanner (EPSON Type 12000XL, EPSON, Meerbusch, Germany) in the photo mode. As document source “transparency attachment” and as template type “color positive film” were chosen. The image type was 8 bit grayscale with a resolution of 600 dpi and a high scan quality. The total root lengths in cm were received by evaluation of the scanned root images using the software WinRhizo Pro 64 bit. The total root length in cm was then converted into the root length density (RLD) per soil layer (cm cm−3).
After the scanning, the roots of each soil layer were dried at 105 °C in a drying chamber for four days and the weight was taken to assess the dry matter yield of roots per soil layer.
For determination of the root/shoot ratio, the root weights of each soil layer were summed up per column, resulting in the root dry matter biomass per column. This was divided by the shoot dry matter yield (shoot + ear dry matter biomass).
Shoot mineral determination
Potassium (K), magnesium (Mg), and phosphorus (P) uptake in shoots were determined with inductively coupled plasma optical emission spectrometry (iCAP 6300 Duo MFC, Firma Thermo Fisher Scientific, Waltham, MA, USA). For this, 5 ml of 65% (v/v) nitric acid and 3 ml of 30% (v/v) hydrogen peroxide were added to 500 mg finely ground (with a vibrating mill: Schwingmühle MM 301, Retsch, Haan, Germany) and dried (at 60 °C) shoot material. The samples were digested in a microwave (Mars 6, CEM, Kamp-Lintfort, Germany) with the following program: 20 min at 200 °C and 2 min at 210 °C and 800 W and 30 min cooling. Finally, the samples were filled up to 50 ml with distilled water and filtrated to collect supernatant (Type 597, Whatman™, VWR International, Radnor, PA, USA). For the determination of total nitrogen (N), 300 mg of the same plant material as used for the determination of K, Mg, and P was weighed in crucibles and assessed by elementary analyses (Vario Max, Elementar, Hanau, Germany).
Shoot water content, transpiration, relative leaf water content, and proline concentrations
For describing shoot water relations, the total shoot water content, the transpiration per column, the relative leaf water content (RWC), and the proline concentrations in leaves were determined. The shoot water content was assessed immediately after harvest of the shoots and was defined as the weight difference of the fresh shoots and of the dried shoots after drying at 105 °C for four days.
The transpiration per column was determined at 59 days after sowing. For this, first the soil surface was covered carefully with cling film in order to exclude water loss via evaporation. Then, starting at 10 am, the weight of each single column was taken. After four hours, a second time the weights of all columns were taken in the same order as in the first measurement. The weight loss during this period was assumed to be the amount of water that the plant lost through transpiration and was calculated as mg water h−1.
The RWC was likewise determined at 59 days after sowing and was performed as described by González and González-Vilar (2001) based on the original method by Barrs and Weatherley (1962). The RWC can be used as indicator for a water deficit in leaves and compares the initial and the turgid water content, on a percentage basis, of leaf disks punched from leaves.
The proline concentrations were determined as described by Woodrow et al. (2017). The fourth youngest leave of a single wheat plant per each column was harvested at 59 days after sowing. Subsequently, the leaves were freeze-dried in a freeze dryer (Christ Beta 1–16, Christ, Osterrode, Germany) for three days. Afterwards, the leaves were ground to a fine powder with help of a vibrating mill (Schwingmühle MM 400, Retsch, Haan, Germany). The proline determination itself was performed in the lab of the Department of Quality of Plant Products of the University of Hohenheim, Germany. First, leaf material was extracted with 70% (v/v) ethanol. The following methodology is based on the fact that proline reacts with the acidified dye ninhydrin and forms a blue-violet to red-brown color complex. The absorption of this complex can be quantified photometrically at a wavelength of 520 nm and with the help of a series of calibrators.
Statistical analysis
The statistical software R (Version 4.0.2, 2020) was used to evaluate the data using the following R packages: “gdata,” “nlme,” “piecewiseSEM,” “multcomp,” “lsmeans,” and “car.” The data evaluation started with the definition of an appropriate statistical model based on generalized least squares. The model included the factors “subsoil moisture,” “pore number column−1,” and “soil layer” as well as all their interaction terms (two-way and three-way). Hereby, the correlations of the measurement values due to the several levels of “soil layer,” if applicable, were taken into account. The residuals were assumed to be normally distributed and to be heteroscedastic with respect to the different levels of “pore number column−1” and “soil layer.” These assumptions are based on a graphical residual analysis. Based on this model, a Pseudo R2 was calculated and an analysis of variances (ANOVA) was conducted, followed by multiple contrast tests (e.g., see Hothorn et al. 2008; Scharschmidt and Vaas 2009) in order to compare the several levels of the influencing factors, respectively.