Soil sampling
Three soil samples with different SOC and nutrient contents were taken from arable fields in North Rhine-Westphalia, Germany (Table 1). Only C3 plants had been cultivated on these fields for at least the last decades, resulting in low δ13C values (Table 1). The altitude of the study sites ranged between 33 and 190 m above sea level. Samples were taken in spring. The fields were cultivated with winter wheat, sugar beet and one was prepared for sowing sugar beet. Basic soil properties are summarized in Table 1. At each sampling site, one sample was taken from 0 to 30 cm depth (Ap horizon) using a spade. The field-moist soil was stored at 7 °C for 1 week and sieved to 2 mm, i.e. all results refer to the fine soil < 2 mm. Sieved soils were divided into three parts: one was dried at 40 °C for chemical analysis, one was stored frozen (− 18 °C) until the start of the incubation experiment, and one was stored field moist at 7 °C for determination of water holding capacity.
Table 1 Properties of the soil samples (arable soils, Ap horizon) Soil physical and chemical analysis
Texture was determined by a combination of wet sieving (sand fraction) and sedimentation (silt and clay fraction) after Köhn (ISO 11277 2002). The total C and N contents of milled soils were determined by elemental analysis (Vario MICRO Cube, Elementar, Hanau, Germany; ISO 10694 1995). All samples were free of carbonates, so total C corresponded to SOC. Plant available P and K were extracted by using the calcium–acetate–lactate method (Schüller 1969; Zbíral 2000). The concentration of P was measured photometrically with the molybdenum blue method (Murphy and Riley 1962) and that of K by atomic absorption spectroscopy (AAS). The pH value was measured in a 0.01 M CaCl2 solution (soil:solution ratio of 1:2.5). Water holding capacity was determined on field moist soil by placing soil into funnels and submerging in water for 30 min. Afterwards, soil was allowed to drain for 24 h, weighed and dried at 105 °C.
Incubation experiment
The sieved and frozen soil samples were defrosted for 2 days at 7 °C. Each soil sample was rewetted or dried to 45% of its water holding capacity, which warrants a sufficient and standardized water availability (ISO/DIS 17155 2001). Each sample was homogenized using a mixer and divided into 30 to 36 plastic vessels, each corresponding to 35 g of dry soil. The samples were then pre-incubated at 22 °C for 96 h to level out effects of mixing and water addition and to stabilize the respiration rate (Blagodatsky et al. 2000).
After pre-incubation, oil was added to the soil samples. To differentiate between CO2 that evolved from the applied oil and CO2 that derived from native SOC, we used commercially produced oil from corn (‘C4 oil’) that differs from the native SOC in terms of its δ13C value (Bol et al. 2003; Kuzyakov and Bol 2006 for distinguishing slurry-derived CO2 from SOC-derived CO2 or Nottingham et al. 2009; Meyer et al. 2017 for distinguishing sugar-derived CO2 from SOC-derived CO2). The δ13C value of the C4 oil was − 16.65‰. The differentiation of oil- and SOC-derived CO2 based on the δ13C value of released CO2 would be straightforward if no isotopic fractionation occurred during mineralization. However, several studies indicated that isotopic fractionation does occur (e.g. Fernandez et al. 2003), which may lead to immense errors in the quantification of CO2 sources. To account for isotopic fractionation of the C4 oil, we followed suggestions of Bol et al. (2003) and conducted the same experiment also with a C3 control. For this, oil with a similar δ13C value as native SOC is required as C3 control. Therefore, we used a C3 sunflower oil, which was mixed with the C4 corn oil at a ratio of 5:1 to obtain the required isotope ratio (− 27.19‰, ‘C3 oil’). Pre-tests revealed no difference in soil respiration rates between the two oils. The δ13C value of oil samples was determined by isotope ratio mass spectroscopy (IRMS, Thermo Delta V Advantage, Thermo Electron, Bremen, Germany) after weighing 0.5 mg of oil and 0.1 to 0.2 mg of C-free ‘Chromosorb’ into tin cups; the latter was used to absorb the oil. Total N contents were determined by elemental analysis as already reported. Total P was determined by ICP-OES (Horiba Jobin Yvon, Ultima 2) after digestion with concentrated nitric acid. Both oils had a C content of 79.6%, a N content of < 0.007% and a P content of 2.9 mg P kg−1 oil (C3 oil) and 3.25 mg P kg−1 (C4 oil), respectively.
The oils were applied with a nozzle according to Peukert et al. (2017). The nozzle height above soil was 15 cm. During application, the oil was mixed constantly with the soil using a vortex shaker. Three different dosages were applied (see Table 2): five drops per vessel, i.e. per 35 g soil (‘low dosage’, 0.1 ml oil vessel−1; ≈ 3 × 10−3 ml g−1 soil), 50 drops per vessel (‘medium dosage’, 1.0 ml oil vessel−1; ≈ 0.03 ml g−1 soil) and 150 drops per vessel (‘high dosage’, 3.0 ml oil vessel−1; ≈ 0.09 ml g−1 soil). The dosages ‘low’ and ‘medium’ were chosen in order to simulate realistic dosages: Peukert (2018) proposed the punctual application of 0.1 ml to maximal 1.0 ml oil to a single weed plant. Thus, assuming that one weed plant grows on the vessel area of 28.3 cm2 the dosages low and medium represent the lower and upper range of recommended dosages. The high dosage of 3.0 ml oil was conducted to simulate effects like inadvertent release of oil or a very high weed cover. The applied oil had a temperature of 22 °C. Note that we also tested the effect of higher oil temperatures during application (100 °C). As results were not significantly different between 22 and 100 °C (Table S1), we concentrate here on reporting only results of the 22 °C treatments.
Table 2 Treatments of the incubation experiment including maximum hourly respiration rate (CO2max) and time until the first respiration peak is reached (tCO2max) (n = 2 or 3) To test effects of nutrient limitation on oil mineralization, the oil dosages low and medium were conducted in combination with an additional nutrient supply. The variant with ‘high’ oil amount was excluded because the stimulation of microorganisms was expected to be too high to allow for a continuous measurement of CO2 release. The applied powdery nutrient mixture contained 47.88 mg (NH4)2SO4 (0.29 mg N g−1 soil), 7.42 mg KH2PO4 (0.048 mg P g−1 soil), and 154.7 mg talcum powder per vessel. These amounts of nutrients were proposed by ISO/DIS 17155 (2001), and talcum powder was used as an inert carrier (Anderson and Domsch 1978). The powder was mixed to the soil immediately before addition of oil using a vortex shaker.
As it is known that microbial nutrient demand forces the acquisition of nutrients from SOM, we tested whether this ‘nutrient mining’ is as effective in providing nutrients required for oil mineralization as the addition of mineral nutrients. To test this, we repeated the abovementioned nutrient addition to selected vessels after finishing the incubation experiment at day 42, i.e. to vessels that already received nutrients at the beginning of the experiment (second nutrient addition) and to vessels that did not receive nutrients at the beginning (first nutrient addition). If nutrient mining by microbes releases the demanded amount of nutrients, CO2 release would not increase upon the late nutrient addition to vessels that had not received nutrients at the beginning.
We also conducted a control treatment, which received neither oil nor nutrients, and a treatment, which received nutrients but no oil. Thus, every soil sample was split into seven treatments with two (for two soils and the dosage low) or three analytic replications each (Table 2).
Soil microbial respiration was measured using an automated respirometer that allows incubating 95 samples in parallel (Respicond VIII, Nordgren Innovations AB, Sweden). The plastic incubation vessels are arranged in a water bath, which ensures a constant soil temperature of 22 °C. The system provides a continuous measurement of CO2 evolution by trapping CO2 in potassium hydroxide (KOH) (Nordgren 1988). Soil respiration was measured hourly for 1008 h, i.e. for 42 days. As the KOH solution has a finite capacity to capture CO2, it was replaced several times during the incubation period in case that about half to three-quarter of the capacity was reached. During replacement, aeration of the samples and equilibration with ambient O2 was allowed.
From the respiration curve, the maximum hourly respiration rate (respiratory peak, CO2max) was derived as an indicator of nutrient availability (Meyer et al. 2017; Nordgren 1992). In case that two respiratory peaks developed during incubation (cf. Fig. 1), CO2max describes the height of the first peak, which usually occurred within the first 130 h after oil addition. Further, we calculated the time needed until this peak was reached (tCO2max), which indicates the viability of microorganisms (ISO/DIS 17155 2001). Total cumulative amounts of released CO2 during the entire incubation period (CO2cum) were calculated by summing up all hourly values of CO2 release.
Quantification of oil- and SOC-derived CO2
To obtain the δ13C value of released CO2, which allows distinguishing SOC-derived from oil-derived CO2, the CO2 captured within KOH had to be precipitated. The KOH solution was replaced several times during the incubation period (see above), and the replaced solution of every vessel was collected in airtight bottles. After completion of the incubation experiment, the bottles (each of them containing the entire KOH collected from each vessel) were shaken and excess BaCl2 solution (3 ml of 1 M BaCl2) was added to a 20 ml aliquot of the replaced KOH. The addition of BaCl2 induced an immediate precipitation of BaCO3. The solution was vacuum-filtered with glass fibre filters and rinsed with distilled water. The remaining BaCO3 was dried at 40 °C and homogenized by grinding. About 1.6 mg of the precipitated BaCO3 was weighed into tin cups, corresponding to 100 μg C per tin cup. Measurements of δ13C were conducted with isotope-ratio mass spectrometry (IRMS, Delta V Advantage Thermo Electron, Bremen, Germany).
The difference in the δ13C value between CO2 respired after additions of C3 oil and C4 oil in each sample, and treatment was used to quantify the proportion of oil- and SOC-derived CO2 to total CO2 release (Bol et al. 2003; Eq. 1). δCO2\C4 and δCO2\C3 are the δ13C values of the CO2 of the respective variant, whereas the subtraction of the C3 variant corrects the microbial isotope fractionation and adjusts this effect. The variables δC4 and δC3 are the δ13C values of the applied C4 oil (δC4) and C3 oil (δC3).
$$ \mathrm{Proportion}\ \mathrm{of}\ \mathrm{oil}\ \mathrm{derived}\ {\mathrm{CO}}_2\ \mathrm{to}\ \mathrm{total}\ {\mathrm{CO}}_2\ \left(\%\right)=\frac{\delta {CO}_{2\backslash C4}-\delta {CO}_{2\backslash C3}}{\delta {C}_4-\delta {C}_3}\times 100 $$
(1)
Absolute amounts of CO2 derived from oil during the whole incubation time were calculated by Eq. 2, i.e. the proportion of oil-derived CO2 from Eq. 1 was multiplied with the respective cumulative amount of CO2 (CO2cum).
$$ \mathrm{Absolute}\kern0.5em \mathrm{amount}\kern0.5em \mathrm{of}\kern0.5em \mathrm{oil}\kern0.5em \mathrm{derived}\kern0.5em {\mathrm{CO}}_2\kern0.5em \left(\mathrm{mg}\kern0.5em {\mathrm{CO}}_2\kern0.5em \mathrm{vessel}-1\right)=\frac{\delta {CO}_{2\setminus C4}-\delta {CO}_{2\setminus C3}}{\delta {C}_4-\delta {C}_3}\times {CO}_{2 cum} $$
(2)
The amount of C applied by the oils depended on the variant. The estimated oil density at 22 °C of 0.91 g ml−1 (following Esteban et al. 2012) was multiplied by the oil amount (ml) of each treatment (oil amount) and the C-proportion of 79.6% of the oils in Eq. 3.
$$ \mathrm{Amount}\ \mathrm{of}\ \mathrm{added}\ \mathrm{C}\ \mathrm{via}\ \mathrm{oil}\ \mathrm{addition}\ \left(\mathrm{mg}\ \mathrm{C}\ {\mathrm{vessel}}^{-^1}\right)=\mathrm{oil}\ \mathrm{amount}\times 0.91\frac{\mathrm{g}}{\mathrm{ml}}\times 0.796\times 1000 $$
(3)
Subsequently, the mineralization of oil was expressed as the loss of oil-C after 42 days of incubation, which was calculated according to Eq. 4, based on a proportion of 27.3% C of CO2. Hence, the CO2 amount derived from oil (Eq. 2) was multiplied by 0.273 and divided by the respective amount of added C (Eq. 3). Thus, we got the relative oil mineralization within the incubation period.
$$ \mathrm{Mineralization}\kern0.5em \mathrm{of}\kern0.5em \mathrm{oil}-\mathrm{C}\kern0.5em \mathrm{after}\kern0.5em 42\kern0.5em \mathrm{days}\kern0.5em \left(\%\right)=\frac{\mathrm{absolute}\kern0.5em \mathrm{amount}\kern0.5em \mathrm{of}\kern0.5em \mathrm{oil}\kern0.5em \mathrm{derived}\kern0.5em {\mathrm{CO}}_2\times 0.273}{\mathrm{amount}\kern0.5em \mathrm{of}\kern0.5em \mathrm{added}\kern0.5em \mathrm{C}}\times 100 $$
(4)
The amount of SOC-derived CO2 was calculated by subtracting the amount of oil-derived CO2 (Eq. 2) from the total amount of CO2 (CO2cum). The priming effect, i.e. the extra amount of CO2 that was derived from SOC compared to the control was calculated according to Eq. 5, i.e. by dividing the amount of CO2 respired from native SOC after oil addition by the amount of CO2 respired in the control (CO2cum Control), which received neither oil nor nutrients or no oil but nutrients (Bol et al. 2003).
$$ \mathrm{Priming}\kern0.5em \mathrm{effect}=\frac{\mathrm{SOC}\kern0.5em \mathrm{derived}\kern0.5em {\mathrm{CO}}_2}{{\mathrm{CO}}_{2\mathrm{cum}}\kern0.5em \mathrm{Control}} $$
(5)
Statistical analyses
The average of the two to three replicates from each sample and treatment was used for statistical analyses. After each replacement of KOH, the values of the first 2 h afterwards were deleted because O2 and CO2 from atmosphere came into the vessels. To investigate differences between the treatments, we conducted a three-way ANOVA with the factors oil dosage, fertilization, and sampling site, whereas the latter was considered as block effect. We also tested the interaction between oil dosage and fertilization. In case of significant effects (p < 0.05), Tukey HSD post-hoc test was carried out to compare differences between treatments. We checked for normal distribution of the data with Shapiro–Wilk test and for variance homogeneity with Levene’s test. However, due to the small number of independent soil samples (n = 3), statistical test results should not be overinterpreted. Yet, they are useful to indicate tendencies. All statistical analyses were performed with R (version 3.2.3, R Core Team 2013).