INTRODUCTION

Soil respiration and its components are under the control of a complex set of biotic and abiotic driving forces, and as croplands are one of the main sources of greenhouse gases into the atmosphere [71] a study of the temporal dynamics of soil respiration has great significance. However, a wide range of studies proved that several factors, like vegetation [5], soil temperature (Ts) [3, 4], soil moisture (SWC) [5, 6], nutrient availability and N treatments [7, 8], and management practices like tillage, harvesting, loosening and sowing [54] can affect soil respiration rates [10–12]. The large uncertainty in Rs estimations could be caused by the fact that Rs is regulated by these multiple biotic and environmental factors [13, 14] and because of the error of measurements [64]. Soil respiration is the second-largest flux after photosynthesis in the global C budget and returns as much as 50–90% of annual gross primary production (GPP) back into the atmosphere [4] depending on these drivers [17, 18]. Combined experiments (field and lab studies) could provide new insights into these effects.

Among these factors, soil temperature and moisture are generally acknowledged as the dominant drivers of Rs but soil temperature is generally considered the most dynamic on both diurnal and longer time scales, therefore it is used in the majority of Rs models [3, 19] being a good predictor of the dynamics of the soil CO2 flux rate. In addition, a strong positive correlation between CO2 efflux and soil temperature was found in natural and agricultural ecosystems [57].

Soil moisture influences the production of CO2 both by directly affecting the activity of microorganism and plant roots and the diffusion of gases through the soil pores [54], and indirectly affecting the change of the substrate supply and plant growth [1, 13, 21]. The changes in soil water content can strongly modify the total soil respiration. Under dry conditions the soil CO2 efflux is low because the activities of micro-organisms and roots are typically low. Increased soil water content normally increases the bio-activity in the soil but if the soil water content is very high the total soil CO2 efflux is reduced and the diffusion of oxygen will be limited resulting in a subsequent suppression of CO2 emission [61].

Rs consists of two main components including autotrophic (Ra) and heterotrophic respiration (Rh) [23–26]. Plants are the most important autotrophs contributing to CO2 efflux from soil by root respiration and heterotrophic respiration from rhizospheric microorganisms (mycorrhizal fungi respiration and the respiration of other root-associated microbes) and is primarily regulated by the root activity and plant photosynthate supply [27, 28]. Although the direct contribution of nematodes and soil macro-fauna (macroscopic invertebrates and small mammals) to Rh is small, they can greatly increase microbial respiration not only by fragmentation and comminution of plant residues [13] but also by predation of some groups of microorganisms [12]. Litter-derived respiration and soil organic matter (SOM) decomposition are considered to be the Rh component [60]. The decomposition depends on substrate availability and soil biota [32–34], the decomposition of SOM was attributed mainly to soil bacteria and fungi and has about 50–55% share in the total soil respiration in dry grasslands [6, 17, 35, 36] reported that Ra was repressed by drought more than Rh in at a grassland site in central Europe highlighting the differences among the components in their response to the environmental variables.

Soil microbial dynamics are controlled through complex interactions with plants and are influenced by a range of organic compounds added to soils from plants as root exudates and as litter inputs [10]. Therefore, the coupling between aboveground gross primary productivity (GPP) and carbon allocation to roots and root-associated organisms varies depending on the season [73], especially the amounts of carbon allocated to the mycorrhizal fungi partners and roots being variable in the different seasons [38, 39]. The droughts typically lead to reduced carbon assimilation in plants [40, 41] and reduced carbon transfer to the roots and the rhizosphere [42, 43], resulting in a lower soil CO2 efflux [31]. Consequently, the reduced belowground carbon allocation weakens plant-microbial interactions [14], as soil microorganisms strongly depend on plant-derived carbon inputs [10].

The main aim of this study was to investigate the temporal dynamics of CO2 efflux from the soil surface in a temperate cropland site during a two-year-long study period, and to analyse the response of the soil respiration components to the main environmental factors such as soil temperature, soil water content, N fertilization and biotic drivers as plant growth. Both field and lab measurements were conducted within the framework of the study.

MATERIALS AND METHODS

Site description. The study was performed from November 2017 to November 2019 in a plot of cropland near Kartal (47.658° N, 19.532° E, 153 m a.s.l.) which is located in the middle part of Hungary and has running eddy-covariance (EC) station since 2017. Gödöllő Experimental Farm Ltd. has the land management rights of the site and provided management data. The average annual temperature was 11.75, 12.94 and 12.91°C and the annual precipitation sum was 620, 552 and 694 mm in 2017, 2018 and 2019, respectively. The soil is chernozem type brown forest soil. Management data contained soil tillage, spraying, sowing, harvesting and fertilizer application timings to the different crop rotations.

Crop rotation of the measured field: 2017 winter wheat, 2018 rapeseed, 2019 sorghum, and 2019–2020 winter wheat. Crops were separated by fallow periods, with no cover crop used between them.

Field design and soil CO2 exchange measurements. The soil CO2 efflux measurements were made from November 2017 to November 2019. Ten PVC rings (10.2 cm of diameter and 5 cm high) were installed one month before the flux measurement to avoid the effects of soil disturbances. PVC rings were inserted approx. 2.5 cm into the ground, leaving 2.5 cm above the ground.

Fluxes of CO2 were measured about bi-weekly/ monthly during a two-year-long study period. PVC rings were left in the field for the entire two-year measurement period, except during the farm operations and they were kept free of any plants for the entire study period.

CO2 efflux was obtained between 10:00 and 12:00 h by closed chamber systems (Licor 6400, LiCor, Inc. Lincoln, NE, USA) in 2017 and 2018 and EGM-4 (PPSystems, Amesbury, USA) in 2019.

Net ecosystem exchange of CO2 (NEE) was measured by eddy-covariance (EC) technique. The EC system at the Kartal site has been measuring the CO2 and sensible and latent heat fluxes continuously since October 2017. It consists of a CSAT3 sonic anemometer (Campbell Scientific, USA) and a Li-7500 (Licor Inc, USA) open-path infra-red gas analyzer (at the height of 2 m, anemometer direction: north), both connected to a CR1000 datalogger (Campbell Scientific, USA) via an SDM (synchronous device for measurement) interface.

Additional data used in the present study included air temperature and relative humidity (HMP35AC, Vaisala, Finland), precipitation (ARG 100 rain gauge, Campbell, UK), volumetric soil moisture content (CS616, Campbell, UK) and soil temperature (105T, Campbell, UK). Fluxes of sensible and latent heat and CO2 were processed by EddyPro® [84] using double rotation, linear detrending and WPL correction [70]. Gap-filling and flux partitioning were performed by the REddyProc online data processing tool [64].

Additional measurements. Soil temperature (Ts) was measured outside the PVC rings concurrently during the soil CO2 efflux measurements in the top 5 cm of soil surface using a thermometer unit attached to the LICOR-6400 or to the EGM-4.

Soil moisture was recorded in the top 7.5 cm, where most of the gas diffusivity from the soil to the atmosphere is likely to occur [43]. SWC was measured by time domain reflectometry (FieldScout TDR300 Soil Moisture Meter, Spectrum Technologies, IL-USA).

Leaf area index was measured by an AccuPar LP-80 ceptometer (Decagon Devices, USA) at each measurement campaign over each plot.

VIgreen (VIgreen index) was derived from red, green, blue (RGB) values of photographs made by a commercial digital camera (Canon Eos 350D) from the measured plots. VIgreen is the normalized difference of reflected green and red light [37]:

$${\text{VIGreen}} = \frac{{{\text{Green}} - {\text{Red}}}}{{{\text{Green}} + {\text{Red}}}},$$
(1)

where VIgreen is a dimensionless index, Green and Red are the component values of a digital image. VIgreen was calculated in R [68].

Bulk density was calculated from the compactness of the topsoil layer measured by a penetrometer (Eijkelkamp, The Netherlands).

Lab measurements. Soil from the top 15 cm layer was collected from the same field and transported to the lab. Before establishing the experiment, the soil was air-dried, visible roots and organic residues were removed and the soil was passed through a 2-mm mesh and then mixed thoroughly; PVC tubes (10.2 cm in diameter and 20 cm height) were used as pots filled up to 15 cm with about 1.6 kg of soil to achieve a bulk density of 1.30 g cm−3. The top 5 cm of the tube was used as a soil respiration chamber during the measurements. Then pots were brought to different soil water content (see below) and were incubated for two weeks before starting the measurement.

The lab experiment was divided into planted soil with maize plants and bare soil. Different soil moisture was set for the measurement series and these moisture levels were binned into two different categories during the data analysis: below 30% (15, 20 and 25%) and above 30% (35 and 40%). Different levels of N fertilizer (N0, N75 and N150) of nitrate ammonium (\({\text{N}}{{{\text{H}}}_{{\text{4}}}}{\text{NO}}_{3}^{ - }\)) were applied on the surface of the soil (Table 1).

Table 1.   Experimental settings of the lab experiment
Full size table

These measurements were conducted in a controlled environment with 12 hours of light, and 20°C of air temperature.

Closed chamber technique was used for measuring the emission of carbon dioxide by a Picarro G1101-i gas analyser. Each sample was measured for 20 minutes, CO2 efflux was calculated using the slope of the concentration change during this period.

CO2 efflux was calculated by the following equation:

$$F = \frac{{n\Delta C~}}{A}~.$$
(2)

Where n represents the number of mols (µmol) in the volume of the closed system, ΔC is the concentration change of the carbon dioxide (µmol mol–1s–1) and A is the area of soil in the PVC tube used in the lab experiment (m2).

Cumulative gas effluxes. The cumulative emissions were calculated using the following formula:

$$T = \sum\limits_{i = 1}^n {({{({{X}_{i}} + {{X}_{{i + 1}}})} \mathord{\left/ {\vphantom {{({{X}_{i}} + {{X}_{{i + 1}}})} {2 \times {{\left( {{{t}_{{i + 1}}} - {{t}_{i}}} \right) \times 24)} \mathord{\left/ {\vphantom {{\left( {{{t}_{{i + 1}}} - {{t}_{i}}} \right) \times 24)} {1000}}} \right. \kern-0em} {1000}},}}} \right. \kern-0em} {2 \times {{\left( {{{t}_{{i + 1}}} - {{t}_{i}}} \right) \times 24)} \mathord{\left/ {\vphantom {{\left( {{{t}_{{i + 1}}} - {{t}_{i}}} \right) \times 24)} {1000}}} \right. \kern-0em} {1000}},}}} ~$$
(3)

where T (g CO2 m−2) is the cumulative CO2 flux, X (mmol CO2 m−2 s−1) is the average daily CO2 flux rate, i is the ith measurement, and (ti+1 − ti) is the number of days between two adjacent measurements.

Data Processing and Modelling. Data processing and statistical analysis were done in R [67]. Gaussian error propagation was used to calculate propagated uncertainties of the cumulative sums and for the averages and model parameters.

Three different soil respiration models were used during the data processing to describe the response of the different CO2 fluxes to the main biotic and abiotic drivers.

In the Lloyd and Taylor model (model 1) soil temperature is the only driving variable:

$$F = a \times {{e}^{{\left( {b\,\left( {\frac{1}{{56.02}}~ - ~\frac{1}{{\left( {{{T}_{{\text{s}}}}\, - \,227.13} \right)}}} \right)} \right)}}},$$
(4)

where F is the soil CO2 efflux (μmol CO2 m−2 s−1), Ts is the soil temperature at 5 cm in Kelvin, a and b are the model parameters.

Model 2 additionally includes SWC [7]:

$$F = a \times {{e}^{{\left( {b\,\left( {\frac{1}{{56.02}} - \frac{1}{{\left( {{{T}_{{\text{s}}}}\, - \,227.13} \right)}}} \right)} \right)\, + \,\left( { - 0.5\, \times \,{{{\left[ {\log \,\left( {\frac{{SWC~}}{c}} \right)} \right]}}^{2}}} \right)}}},$$
(5)

where SWC is the volumetric soil water content (%) and c is a model parameter.

Model 3 is extended model 2 by adding VIgreen as a driving variable:

$$F\, = a \times {{e}^{{\left( {\left( {d\, \times \,VIgreen} \right)\, + \,b\,\left( {\frac{1}{{56.02}} - \frac{1}{{\left( {{{T}_{{\text{s}}}} - \,227.13} \right)}}} \right)} \right)\, + \,\left( { - 0.5\, \times \,{{{\left[ {\log \left( {\frac{{SWC}}{c}} \right)} \right]}}^{2}}} \right)}}},$$
(6)

where VIgreen is the Vegetation index and d is a model parameter.

RESULTS AND DISCUSSION

Field Experiment

Meteorological and environmental conditions during the study period. The maximum Ts (38.5°C) were observed in June 2019 and the highest SWC (57.6%) was observed in November 2019. The minimum Ts (1.11°C) was observed in December 2017 and the lowest SWC (4.1%) in January 2019, while the maximum Ta (34.57°C) was observed on 12th August 2019 and the minimum Ta (–11.56°C) was observed on 28th February 2018.

The annual sum of precipitation in 2018 was lower (552 mm) than it was in 2017 (620 mm), while it was the highest in 2019 (694 mm).

The values of the VIgreen measured during 2018–2019 varied between –0.062–0.342 and –0.064–0.259, respectively. It was lower than 0 when no vegetation was present in the field (fallow periods), while it was rapidly growing after sowing and germination. The highest VIgreen values were related to the peak green biomass of the crops, observed on 16 April 2018 in wheat (0.342) and 26 June 2019 in sorghum (0.244) (Fig. 1, middle panel).

Fig. 1.
figure 1

Top panel: Seasonal variations of soil moisture (SWC, %, blue dots) in the 0–7.5 cm soil layer, 5 cm depth soil temperature (Ts, °C, red dots). Middle panel: seasonal variations of leaf area index (LAI, m2 m–2, brown dots) and VIgreen index (VIgreen, green dots). Lower panel: crop rotation of the measured field (Winter wheat, Rapeseed, Sorghum), and soil respiration (Rs, whiskers showing standard deviation) during the two-year-long study period. Downward-facing arrows indicate the timing of sowing and harvesting in the site.

Full size image

While the values of LAI were equal to 0 when no vegetation was present in the field, the highest leaf area index was observed during the last stages of crop growth, on 16 May 2018 it was 5.005 m2 m–2 and on 15 August 2019, it was 5.653 m2 m–2 (Fig. 1, middle panel).

Seasonal variation of soil respiration. Soil respiration values were low during winter, increased in spring and reached their maximum during the summer periods of 2018 and 2019 and started to decrease at the beginning of autumn (Fig. 1). The highest emissions of 7.04 ± 0.44 µmol CO2 m–2 s–1 were detected immediately after soil loosening in the fallow period in August 2018 at an intermediate soil water content of 26% and soil temperature of 23°C.

Soil respiration is typically related to air or soil temperature, soil water content and in more recent cases to substrate supply [3, 52]. At the tillering stage the air and soil temperature gradually increases, plants grow quickly, soil microbial activities are enhanced and root exudate production increases, providing suitable conditions for soil respiration [79]. Furthermore, [74] concluded that intermediate soil moisture conditions (between 20 and 60% WFPS) produced the highest CO2 emissions. Photosynthesis was also proposed as one of the controlling variables in soil respiration [27, 55].

The second-highest emission of 5.72 ± 3.72 µmol CO2 m–2 s–1 was observed in June 2019 a few weeks after sorghum sowing and N fertilizer application, accompanied by higher soil water contents (42%) due to a heavy rainfall before the day of the measurement. The ample water availability in the soil, plant activity (VIgreen, 0.256) and high soil temperature (29°C) all resulted in a peak in soil CO2 emission rate (Fig. 1). According to previous studies the impacts of N addition on CO2 efflux varied widely with the level of N addition resulting in contradictory viewpoints concerning whether N applied to soils (regardless of its forms) increases soil CO2 production or not [56, 57]. In addition, a previous study suggested that increased N supply significantly stimulated CO2 emission and these conditions generally promoted autotrophic plant respiration of above- and belowground parts [18], as well as rhizosphere respiration by microbes due to the accelerated decomposition of soil organic matter [65]. Therefore, these conditions are suitable for greater root respiration and more priming for the microbes [61].

Soil respiration decreased by 0.17 ± 0.006 µmol CO2 m–2 s–1 in November 2018. Accompanied by 19% of SWC and 5°C of soil temperature, this lower efflux was due to the lack of vegetation in the field because the sowing of rapeseed at the beginning of autumn in 2018 wasn’t successful. Another possible reason was the low temperature [76]. However, temperature and plant biomass were good proxies for variations in both autotrophic and heterotrophic capacity for soil respiration [61, 60]. While it decreased substantially in the winter by 0.06 ± 0.007 µmol CO2 m–2 s–1 in February 2019 with 20% of SWC and at a temperature of 3°C (Fig. 1). This lower efflux was due to the low temperature and to the fact that the autotrophic respiration was generally very low or zero because there was no vegetation growing in the study site. However, the heterotrophic respiration (soil microorganisms: bacteria and fungi and other micro-organisms) could maintain both catabolic (CO2 production) and anabolic processes (biomass synthesis) under frozen conditions [32]. Thus, a gaseous exchange between the atmosphere and soil does not stop even in frozen soil, resulting in the accumulation of CO2 during winter and its release into the atmosphere during spring thaw events [15].

Soil respiration showed a positive correlation with soil temperature R = 0.57, but no other investigated variable showed a significant correlation with soil respiration (Fig. 2). Soil temperature was found to be the principal factor influencing soil respiration on both diurnal and longer time scales [5], it is used in the majority of Rs models [3, 19, 55] due to its general effect on soil microclimate conditions and the biological activity of below-ground organisms [64, 65]. The eventual influence on soil respiration by the variation of soil temperature as observed in the present study was similar to previous studies [9, 67, 75]. Using an exponential model (Model 1, [56], Eq. (4)) between CO2 efflux and soil temperature, the goodness-of-fit was r2 = 0.4 (Table 2). Since there is no one single widely accepted model type that can describe the relationship between soil CO2 efflux and soil water content [23], using Lloyd and Taylor soil respiration model extended by a log-normal function of soil water content (Eq. (5)) would allow to include SWC in the modelling and this way the goodness-of-fit value had slightly improved (r2 = 0.45). Furthermore, using soil respiration model extended by a log-normal function of soil water content and by an exponential function of VIgreen (Eq. (6)) was apt to represent better the response of soil respiration to these factors at our site with r2 = 0.54 (Table 2).

Fig. 2.
figure 2

Correlation plot between soil respiration and SWC (soil water content), VIgreen (VIgreen index), LAI (leaf area index), Ts (soil temperature), BD (bulk density of the soil) and NEE (net ecosystem exchange of CO2). Only statistically significant (p < 0.05) correlations are presented.

Full size image
Table 2.   r2 values, and model coefficients for model 1, 2 and 3. Statistical significance levels of the coefficients and model fitting were p-value <0.0001 in all cases
Full size table

Fitted parameters of the three soil respiration models (model 1, 2 and 3, Table 2) show that Model (3) where Ts, SWC and VIgreen were included was the best fit because the r squared value improved with the increasing number of variables. The log-normal shape of soil moisture-respiration response was proposed before [22, 52]. It originated from the Michaelis Menten kinetics of the response of respiration to substrate and oxygen availability [25].

The reflected green and red lights of the surface obtained by commercial digital camera (Canin Eos 350D) change with the different phenological stages of the vegetation during the seasons [63]. [62] conducted an experiment to prove that VIgreen will change according to the season, therefore it can be incorporated into soil respiration models [45].

Lab Experiment

Cumulative CO2 efflux course with different levels of N treatment in the presence/absence of plants. Lab measurements were aimed at quantifying the effect of the presence/absence of plants, the effect of soil moisture and the effect of different N addition (0, 75, and 150 kg N ha–1) on the cumulative CO2 efflux.

The data implied that more than three weeks after N fertilization (on days 21 after fertilization), the cumulative CO2 efflux in bare soil at N75, N150 were 0.63 ± 0.01 g CO2 m–2 and 0.90 ± 0.02 g CO2 m–2, respectively (Fig. 3, left panel), which is higher than that in the N0 treatment (60 ± 0.02 g CO2 m–2). However, much higher cumulative CO2 effluxes were observed in planted soil samples in all treatments N0, N75 and N150: 1.05 ± 0.02 g CO2 m–2, 1.26 ± 0.03 g CO2 m–2 and 1.30 ± 0.03 g CO2 m–2, respectively.

Fig. 3.
figure 3

Cumulative CO2 efflux (g CO2 m–2) courses across the 3 weeks long laboratory study period. CO2 effluxes are separated by N treatments (left panel, 0, 75 and 150 kg N ha–1) and by soil water content (right panel, >30 and <30%) in bare and planted soil.

Full size image

The efflux values from planted soil N0, N75 and N150 were around twice as high as N0, N75 and N150 in bare soil, respectively. The difference between planted and bare soil in our study was due to the activity of plants resulting in root respiration and the priming effects of root exudates on soil microbes [52], which, in turn, improved soil nutrient content [58, 73], and accelerated the decomposition of soil organic matter [65]. The cumulative CO2 emissions in our present study was found to have a positive correlation with the stand age of the plant and with N fertilizer rates: as the plants grew and more N was added more CO2 was emitted. A previous study suggested that N addition stimulated CO2 emission by promoting autotrophic plant respiration (above and below ground parts) [18] as well as heterotrophic respiration (rhizospheric respiration) by microbes due to the accelerated decomposition of SOM which was discussed above. Other studies also found a significant increase in soil respiration in unplanted and N fertilized soil compared to the unfertilized soil [74–76].

Relationship between the cumulative CO2 efflux and soil water content. Pearson’s correlation of CO2 efflux and soil moisture indicated that soil moisture was well correlated with CO2 emission (R = 0.43). Figure 3 right panel shows that the cumulative CO2 efflux increased with increasing SWC, the efflux was significantly (almost three times) higher (by as much as 2.3 ± 0.05 g CO2 m–2) in planted soils at higher soil moisture levels (>30% and after three weeks of N fertilization) than at the lower soil moisture levels (<30%) by 0.92 ± 0.01 g CO2 m–2, and, similarly, three times higher than in bare soil at higher SWC by 0.86 ± 0.02 g CO2 m–2. Therefore, the effects of plant presence and soil moisture on soil respiration had similar magnitude.

In bare soil the cumulative CO2 efflux was also significantly lower (by as much as 0.59 ± 0.03 g CO2 m–2) at the lower soil moisture level (<30%) than at higher soil moisture level (>30%) (by 0.86 ± 0.02 g CO2 m–2 (Fig. 3 right panel). Similarly to our results, increased CO2 fluxes were observed in soils with higher SWC under maize in other studies as well [77, 78]. This higher efflux at higher soil water content in the presence of plants indicate greater root and rhizosphere respiration and increased SOM decomposition and thus more CO2 emission [61].

The influence of moisture content on soil CO2 efflux is complex through its effect on respiratory activity of roots and microbes [80] and gas transport through the soil [33]. Generally, soil CO2 efflux increases as soil moisture increases but soil moisture content can significantly reduce soil CO2 efflux at its highest (wet soil) by blocking CO2 transport because of low soil effective porosity [5, 81] and at its lowest (dry soil) [10, 82, 83] by limiting respiration substrate availability and thereby it reduces soil respiration [27, 84, 85, 86]. However, soil respiration is more responsive to the combined effect of soil water content and soil temperature [87, 88]. In our study the higher soil moisture levels (35 and 40%) could enhance respiration rates and no negative effect of high soil moisture was observed.

CONCLUSIONS

Field and laboratory experiments were performed during a two-year-long study period (from November 2017 to November 2019) to quantify the different effects of principal biotic and abiotic drivers on soil CO2 efflux. We found that the highest CO2 emission rates occurred during summer and the lowest rates during the snow-covered winter period, and that soil temperature, soil water content, agricultural management practices and plant growth were the principal drivers playing a major role in the carbon cycle at this temperate cropland site. We aimed to separate the effect of these drivers on CO2 efflux in our laboratory study and we found that the cumulative CO2 efflux in the N application was higher than that it was in zero-N treatment in both planted and bare soil, therefore the presence of plants and their growth can explain the temporal variations in cumulative CO2 efflux due to root biomass. On the other hand, significant positive correlations between CO2 efflux and soil moisture were found, indicating that soil water content was the main factor limiting the rate of the CO2 emission from the soil in our laboratory study.