Soil heterotrophic and autotrophic respiration respond differently to seasonal variations in temperature and water content under monsoon continental climate

Abstract

Soil respiration (R_S) comprises terrestrial ecosystems’ second-largest carbon flux. Yet, methodological errors in R_S partitioning and uncertainties in seasonal responses of R_S make it difficult to predict future R_S. Here, we tested the assumption of R_S partitioning (similar microbial respiration between planted and root-free soils), and explored two components of R_S, autotrophic and heterotrophic respiration (R_A, R_H, respectively), in a temperate grassland under monsoon continental climate. Microbial respiration in soils from planted plots was 3.88 times higher than that from root-free plots during lab incubation. In field, R_H:R_S ratio was relatively low during non-monsoon, but increased during monsoon. The R_H was more sensitive to temperature than R_S, indicating a greater Q₁₀ of R_H than that of estimated R_A. The annual R_H:R_S excluding the monsoon period was comparable to those reported in the global Soil Respiration Database (SRDB) and other Korean literature. This study highlights that the assumption of R_S partitioning can be violated, that R_H exhibits a greater sensitivity to changes in temperature and soil water content than R_A, and that annual R_H:R_S may be similar across the globe when extreme precipitation (e.g., monsoon) is excluded.

Introduction

Soil respiration (R_S) is the CO₂ release from soil to the atmosphere via autotrophic respiration (R_A, roots and mycorrhizal fungi) and heterotrophic respiration (R_H, saprotrophic microbes)^1,2. On a global scale, R_S releases approximately 110 Pg C to the atmosphere annually^3,4, representing the second largest carbon (C) flux in terrestrial ecosystems⁵. Despite the substantial magnitude of R_S the seasonal variability of R_S makes it difficult to accurately predict R_S^6,7,8. This is because we still have limited knowledge about how R_S will respond to changes in soil temperature and water content, a common condition especially in a monsoon climate, and how the partitioning of R_S into R_A and R_H will change with seasons^9,10. Given that 19.4% of Earth’s surface exhibits monsoon climates and that extreme temperature and precipitation regimes are expected to increase under future climate scenarios^11,12, it is necessary to explore R_A, R_H, and R_S with changes in temperature and water content to improve the accuracy of future climate projections.

Soil temperature and water content can differentially influence R_A, R_H, and R_S^2,13. While increasing temperature enhances the rates of biochemical reactions, temperature sensitivity (Q₁₀; the change in the reaction rate with a 10 °C increase in temperature) can vary with the types of reactions. For example, Q₁₀ of R_H is reported from 1.43 to 5.48^14,15, while that of R_S is between 1.07 and 6.60¹⁶. Different biochemical reactions involved in R_A and R_H can lead to the differences in the Q₁₀. R_A is influenced by plant phenology and the photosynthate allocation to roots^17,18. In contrast, R_H is influenced by substrate availability, microbial physiological responses, microbial C demand, and community composition¹⁹.

Soil water content can also influence R_H and R_S by altering substrate availability and microbial physiological responses^20,21,22. At relatively low soil water content, microbial metabolic activity can be restricted due to reduced diffusion of organic C (OC)²³. When precipitation falls, relatively high soil water content can induce CO₂ pulses from soils, via OC dissolution and diffusion, promotion of the release of cellular metabolites, and stimulation of microbial activity^24,25. As such, shifts in OC availability and microbial physiological responses upon changes in soil water content should be accounted for when exploring R_S in regions with huge fluctuations in rainfall over seasons. Specifically, Korea’s climate is classified as a monsoon continental climate, with heavy rainfall occurring in summer. The summer monsoon can lead to rapid increases in soil water content and trigger the CO₂ pulses from soils, known as the “birch effect”^26,27,28. So far, only few studies have explored how the partitioning of R_S into R_A and R_H will change with rapid changes in soil water content^29,30. For example, R_A responds to heavy rainfall more slowly than R_H²⁹, partly because a suite of conditions, such as CO₂ level and light intensity, can also influence R_A as well as water availability³¹.

Root exclusion is commonly used to partition R_S into R_A and R_H^32,33. This approach either uses bare plots (bare soil method) or cuts live roots around plots (trenching method), to prevent roots from respiring newly produced photosynthates^33,34. Neither the bare soil nor trenching method determines R_A directly. Instead, both methods estimate R_A by subtracting R_H from R_S³⁵. The assumption behind this calculation is that microbial respiration from the root-free plots is the same as microbial respiration from the intact, planted plots³⁶. However, this assumption has rarely been explicitly tested, raising concerns about the accuracy of R_A and R_H estimation. The presence and activity of roots often stimulate microbial respiration and growth in the rhizosphere³⁷. If this holds true, microbial respiration from the root-free plots should be lower than that from planted plots, therefore decreasing the accuracy of R_A estimation.

Here, we explore how seasonal temperature and water content variations influence R_H and R_S in a grassland under a monsoon continental climate, a relatively less studied ecosystem and climate, to obtain a more detailed R_S assessment. Also, we compared our R_H:R_S ratio to those from the global Soil Respiration Database (SRDB)³⁸ and domestic forest data (published in Korean, so not included in the SRDB) to evaluate the generalizability of our results and to better constrain regional soil C dynamics. Finally, we tested the assumption of root exclusion methods that microbial respiration is similar between root-free and planted plots. We hypothesized that (1) microbial respiration will be higher in planted plots than in root-free plots, (2) R_H will respond to seasonal variations in temperature and water content to a greater degree than R_A, and (3) the average R_H:R_S ratio in this study will differ from those reported worldwide in the SRDB.

Results

Initial soil physicochemical and microbial properties

Before the field measurement of R_S and R_H, we characterized initial soil physicochemical and biological properties in the M. × giganteus and M. × giganteus-free soils. Soil physicochemical characteristics were similar between the M. × giganteus and M. × giganteus-free soils, except for sand and silt contents (Table 1). The K₂SO₄-extractable OC tended to be greater in the M. × giganteus soils compared to the M. × giganteus-free soils (p = 0.114).

Table 1 Initial soil physicochemical and biological characteristics in the M. × giganteus and M. × giganteus-free soils (mean ± standard error) (n = 4).

The M. × giganteus soils harbored more root biomass and microbial biomass C compared to M. × giganteus-free soils (Table 1). Microbial respiration assessed in the laboratory was 287.50% higher in the M. × giganteus soils compared to the M. × giganteus-free soils (p < 0.001, Fig. 1). Microbial respiration was significantly correlated with microbial biomass C and K₂SO₄-extractable OC (Pearson correlation coefficient r = 0.71, p < 0.001 for microbial biomass C; r = 0.74, p = 0.04 for K₂SO₄-extractable OC).

Fig. 1

Microbial respiration from the M. × giganteus soils and M. × giganteus-free soils in laboratory. Error bars indicate the standard error of the mean (n = 4).

Seasonal soil temperature, water content, R_H, and R_S

We assessed soil temperature, soil volumetric water content, R_H and R_S between May and December in 2023. Soil temperature was similar between the M. × giganteus and M. × giganteus-free plots over the whole study period (Fig. 2a). The summer monsoon occurred from June 25th to July 26th, when soils received 50% of the annual precipitation (179.80% of the 30-year average monsoon period rainfall)³⁹. Soil water content was significantly higher in the M. × giganteus plots than the M. × giganteus-free plots during the summer monsoon (p = 0.001), but was similar in other periods (Fig. 2b). Both R_H and R_S exhibited significant seasonal variability (p < 0.001 for both), with the monthly average R_H:R_S ratio ranging from 0.35 to 1.12 (Table 2). While R_H was lower than R_S during the non-monsoon period, R_H exceeded R_S during the monsoon period when soil water content was relatively low in the M. × giganteus-free plots (Fig. 3 and Fig. 2b).

Fig. 2

Seasonal variations of (a) soil temperature and (b) soil volumetric water content at a depth of 6 cm, and precipitation in M. × giganteus and M. × giganteus-free plots from April 2023 to December 2023. The shaded area represents the monsoon period from June 25th to July 26th, 2023. Error bars indicate the standard error of the mean (n = 4).

Table 2 Monthly averages of R_H (heterotrophic respiration), R_S (soil respiration), and R_H:R_S in the study site (mean ± standard error) (n = 4). The R_H:R_S values without standard errors represent data measured only once per month.

Fig. 3

Seasonal variation of soil respiration (R_S) and heterotrophic respiration (R_H) in the M. × giganteus and M. × giganteus-free plots over the study period. Error bars indicate the standard error of the mean (n = 4). The shaded area represents the monsoon period from June 25th to July 26th, 2023.

To clarify the seasonal variability of R_H and R_S, we performed a nonlinear model analysis with CO₂ flux as a dependent variable, and temperature and water content as explanatory variables (Table 3). During the non-monsoon period, both soil temperature and water content significantly influenced R_H and R_S (Table 3, Fig. 4a and 4b). In contrast, we did not find significant effects of temperature and water content on R_H and R_S during the monsoon period (Table 3, Fig. 4c and 4d). When we plotted CO₂ flux as a function of soil temperature, R_S and R_H were relatively well explained by soil temperature (R² = 0.81 and 0.54, respectively), with Q₁₀ values of 2.29 and 2.67 for R_S and R_H (Fig. 5).

Table 3 Relationships between soil CO₂ flux (μg C cm^-2 h⁻¹), soil volumetric water content (cm³ cm⁻³), and soil temperature (°C) .

Fig. 4

Response surface showing the relationships between soil temperature (°C) and soil volumetric water content (m³ m⁻³) with: (a) soil respiration (R_S) during non-monsoon period, (b) heterotrophic respiration (R_H) during non-monsoon period, (c) R_S during monsoon period, and (d) R_H during monsoon period. Observed values (black dots) represent measured soil CO₂ flux, and modeled values (colored surfaces) are predictions derived from fitted to the observations.

Fig. 5

The relationships between soil CO₂ flux and soil temperature (T, °C) in the M. × giganteus and M. × giganteus-free plots using observed whole data over the non-monsoon period. The red and blue dashed lines indicate the exponential curves of R_H (heterotrophic respiration) and R_S (soil respiration) with soil temperature, respectively.

Comparison of R_H:R_S ratio

We compared our R_H:R_S ratio to those reported in Korean literature and the SRDB-V5 (Fig. 6 and Table 4). The average R_H:R_S ratio during the non-monsoon in this study was comparable to the average R_H:R_S ratio from only one grassland study in Korea (Table 4) and from other ecosystems worldwide (SRDB-V5 data, Fig. 6). In contrast, the average R_H:R_S ratio during the entire period including the monsoon was higher than the R_H:R_S ratio reported in the SRDB-V5 (Fig. 6).

Fig. 6

The ratio of heterotrophic respiration to soil respiration across ecosystem types. Data are drawn from a global Soil Respiration Database (SRDB) and our study site. Linear regression lines were fitted to each ecosystem type, with forest in solid green (y = 0.51x + 47.19, R² = 0.62), grassland in solid blue (y = 0.47x + 60.98, R² = 0.71), agriculture in solid purple (y = 0.42x + 78.78, R² = 0.37), this study during the non-monsoon period in solid pink (y = 0.79x – 267.13, R² = 0.68), and this study during the entire period in dashed pink (y = 1.04x – 333.83, R² = 0.71).

Table 4 Comparison of R_S and R_H (g C m⁻² yr⁻¹) among ecosystems based on the global Soil Respiration Database (SRDB) and Korean data (mean ± standard error).

Discussion

Laboratory measurement of microbial respiration in the M. × giganteus soils and M. × giganteus-free soils

Consistent with our first hypothesis, microbial respiration was higher in the M. × giganteus soils compared to the M. × giganteus-free soils (Fig. 1). Greater microbial biomass C and respiration in the M. × giganteus soils (Fig. 1 and Table 1) suggest that roots have stimulated soil microbial activities. Living roots release approximately 5–21% of the total C fixed through photosynthesis via root exudates^40,41. This release of easily available OC likely contributes to the enhanced microbial activities in M. × giganteus soils. Indeed, we observed relatively higher concentration of K₂SO₄-extractable OC in the M. × giganteus soils (Table 1). Dissolved organic carbon (DOC), a primary C substrate for microorganisms^42,43, is positively correlated with microbial respiration (R_H)⁴³. In line with these studies, we observed a significant positive correlation between K₂SO₄-extractable OC and microbial respiration (p = 0.04). Thus, our results imply that microbial communities in the M. × giganteus soil had greater access to easily available OC than those in the M. × giganteus-free soils, and that the presence of M. × giganteus roots likely stimulates microbial activity via root exudates.

Our findings that microbial respiration differed between plots with or without M. × giganteus have implications for the interpretation of R_H and R_A in R_S partitioning. Traditionally, root exclusion methods assume that microbial respiration from planted and root-free soils is identical³³. Yet, some studies reported that the assumption can be violated if the decomposition of dead fine roots in trenched soils causes an overestimation in R_H^32,34. While our finding is similar to these studies in that microbial respiration differs between planted and root-free soils, we demonstrate that R_H is likely to be underestimated, not overestimated, in the bare soil method. This is because the bare soil method did not involve cutting live roots. Some dead roots found in these plots were likely debris from long-deceased plants that had previously occupied the M. × giganteus-free plots, and therefore cannot promote microbial respiration (no overestimation of R_H). Previous studies report that R_S partitioning method can influence the estimation of R_H and R_A, potentially introducing biases in CO₂ flux calculations^34,44. This along with our results highlights the need to interpret results within the context of methodological choices in R_S partitioning, to improve the accuracy of soil CO₂ flux estimation and better predict ecosystem C dynamics.

We acknowledge that we measured microbial respiration once in the laboratory in mid-June, and that we do not know how much microbial respiration would vary with seasons. However, the peak season of plant growth is from July to August, during which rapidly growing plants release more root exudation and thus more DOC^45,46. Given that our sampling was conducted in mid-June when root exudates might have been less abundant compared to the peak of plant growth, soils sampled during this period may have lower microbial respiration than those sampled in July and August. This suggests that microbial respiration may be underestimated to a greater extent than we proposed.

Field R_H and R_S over seasons

Our results supported the second hypothesis that R_H is more sensitive to seasonal variations in temperature and water content than R_A. Relatively high Q₁₀ of R_H and the R_H:R_S ratio (Fig. 5 and Table 2) indicate greater responses of R_H than R_A to seasons. The greater seasonal response of R_H than R_A can be explained in two ways. First, respiratory substrates differ between R_A and R_H. In the M. × giganteus-free plots, labile C inputs (i.e., fresh litter, root exudates) are absent, prompting microbes to respire relatively old and complex C compounds. In contrast, roots respire fresh carbohydrates derived from photosynthesis^47,48. The Arrhenius equation dictates that complex compounds require greater activation energy for the reaction to proceed, resulting in higher Q₁₀ values^49,50,51. As such, R_H derived from relatively old and complex C compounds can exhibit a greater Q₁₀ than R_A.

Second, while temperature and water are the two most important factors for heterotrophic respiration, other environmental factors such as radiation intensity and nutrient availability can also affect autotrophic respiration. High temperature and water availability during the summer monsoon greatly enhanced heterotrophic respiration (Fig. 3 and Table 2). Yet, autotrophic respiration may have suffered from a lack of radiation and associated reductions in photosynthesis^31,52,53. In fact, a more rapid increase in R_H with rising temperature compared to R_A has been observed in other studies when water is not limiting^29,53,54.

Interestingly, during the monsoon period, the R_H was higher than R_S (Fig. 3) and soil water content was lower in the M. × giganteus-free plots compared to the M. × giganteus plots (Fig. 2b). These observations deviate from the common expectation: higher R_S than R_H and greater soil water content in root-free soils due to no plant water uptake³². We ascribe our observations to the litter accumulation and saturation of soil pores. In the M. × giganteus plots, the dead shoots from previous years have accumulated on soil surface, likely decreasing evaporation from soil to the atmosphere. Also, we frequently observed the formation of puddles in the M. × giganteus plots during the monsoon. This was because the accumulated litterfalls and overshooting rhizomes promoted microtopography conducive to waterlogging. In contrast, heavy rainfall during the monsoon may have directly infiltrated soil in the M. × giganteus-free plots, facilitating the diffusion of soluble C and the release of intracellular osmolytes⁵⁵. This influx of readily available substrates likely stimulated microbial activity, fueling the observed pulse of R_H in M. × giganteus-free plots. Combined, litter accumulation and waterlogging seemed to inhibit O₂ diffusion and R_S, while rapid rainfall infiltration facilitated substrate availability and enhanced R_H, ultimately causing R_H to exceed R_S during the monsoon.

Comparisons of R_S and R_H among ecosystems

Our third hypothesis that the R_H:R_S ratio will differ between our data and the SRDB was partly supported. During monsoon, the R_H:R_S ratio in this study was greater than the world R_H:R_S ratio (Fig. 6). Yet, during non-monsoon, the R_H:R_S ratios were similar among literatures. For example, the annual mean R_H:R_S was 0.59 in forests, 0.57 in grasslands, and 0.62 in agriculture ratio in the SRDB, which were similar to the R_H:R_S ratio from Korean literature (Table 4). Similar R_H:R_S ratio across ecosystems and between Korea and world data imply that it may be feasible to use global R_H:R_S ratio to fill the ecosystem gap with little data. Furthermore, the global similarity of R_H:R_S ratio when extreme precipitation is excluded highlights the importance of considering rainfall variability of monsoon climate in soil C dynamics. These findings indicate that extreme monsoon rainfall significantly influences the R_H:R_S ratio, and that neglecting this variability may lead to biases in soil C dynamics modeling in monsoon regions. Therefore, integrating extreme rainfall variability into predictive models is essential for accurately quantifying soil CO₂ fluxes and better capturing the spatial and temporal patterns of the R_H:R_S ratio.

Conclusions

We demonstrate significant variability in R_S, R_H, and the R_H:R_S ratio across seasons under continental monsoon climate. Our results highlight that R_H is more sensitive to changes in temperature and soil water content than R_A. Also, our observation of lower microbial respiration in root-free soils emphasizes the need to carefully consider R_S partitioning methods for accurately estimating and predicting CO₂ flux from plants and soil in terrestrial ecosystems. Furthermore, relatively similar R_H:R_S ratios during non-monsoon across ecosystems suggest the possibility of employing world data to estimate R_H from less-studied ecosystems or regions. Notably, the significant increase in R_H:R_S during the monsoon highlights the strong influence of extreme rainfall events on soil respiration dynamics. Further studies are required to gain a deeper mechanistic understanding of the long-term or cumulative effects of extreme summer monsoon rainfall and temperature in a monsoon continental climate across various ecosystems. Additionally, to improve the accuracy of C flux estimations, predictive models should account for biases in R_H and R_A arising from different partitioning methods and consider extreme rainfall variability. Incorporating these factors will enable more precise C flux predictions and facilitate a more consistent observation of global R_H:R_S ratio trends.

Materials and methods

Site description

This study was conducted at the Experimental Farm of Seoul National University, Suwon, Korea (37°16′ 12.598" N, 126° 59′ 24.529" E). The region experiences a monsoon continental climate, with four distinct seasons including hot and humid summers and cold winters. The 10-year mean annual temperature is 13.04 °C, with the minimum daily mean temperature of −6.7 °C in January and a maximum of 31.2 °C in August. The mean annual precipitation is 1,320 mm, falling primarily between June and July during the summer monsoon period (https://data.kma.go.kr/cmmn/main.do).

In 2011, twenty Miscanthus × giganteus plots were established to study plant-soil C interactions and use M. × giganteus as biofuel production. We selected M. × giganteus due to its extensive rhizome and root biomasses, which drive belowground C cycling⁵⁶. The M. × giganteus plots are part of a long-term Nitrogen (N) addition with 0, 30, 60, 120, and 240 kg N ha⁻¹ year⁻¹ (total 20 plots = 5 N treatments × 4 replicates, 2 m × 4 m each). Each plot was housed in a 1.64 m deep lysimeter and separated from one another. Among the M. × giganteus plots, we only used the control plots (0 kg N ha⁻¹ year⁻¹) for this study.

The bare soil method was used to partition R_S into R_A and R_H. We leveraged the M. × giganteus plots in a separate lysimeter to measure R_S and considered M. × giganteus-free soils outside the lysimeter as bare, root-excluded plots (hereafter M. × giganteus-free plots, n = 4) to measure R_H. We regularly removed any plants growing in the M. × giganteus-free plots by pulling them out to minimize root respiration.

Soil sampling and initial physicochemical properties

In June 2023, we collected surface soils (0–10 cm depth) from the M. × giganteus plots and M. × giganteus-free plots using a Gas-powered core sampler (AMS Inc., USA) with a PVC tube liner (inner diameter: 3.81 cm) and a soil auger (0–6 cm depth), respectively. A total of 8 soil cores (2 treatments * 4 replicates) were promptly stored in a cooler at 4 °C, passed through a 2 mm sieve, and transferred to the laboratory at Seoul National University. We collected roots larger than 2 mm from each soil core, oven-dried at 55 °C for 2 days, and recorded the weights. Soil pH was determined on 1:5 (w/v) soil:water slurry (S220, Mettler Toledo, Switzerland). Gravimetric water content was assessed by drying 5 g soils at 105 °C for 48 h. We assessed water holding capacity by saturating 10 g of soil with 25 mL of deionized water and draining it for 4 h using a filter funnel (Qualitative filter papers no.1, Whatman™, UK). Total C and N were measured on sieved, 105 °C dried, finely-ground soils at the National Instrumentation Center for Environmental Management (NICEM), Seoul National University. K₂SO₄-extractable OC was quantified by mixing a 5 g soil sample with 25 mL of 0.5 M K₂SO₄ and shaking the mixture at 200 rpm for 4 h. Subsequently, the slurries were centrifuged at 1,000 × g for 5 min (MF-600 plus, Hanil, Korea), and then filtered using 0.45 μm pore-size vacuum filtration (Sartorius, Göttingen, Germany). The total amount of OC in the K₂SO₄ extracts was then analyzed using a total OC analyzer (TOC-L, SHIMADZU, Japan) at the NICEM. Bulk density was calculated as the dry weight of soil divided by its total soil sample volume after correcting rock volume and weight. We determined particle size distribution using the hydrometer method.

Lab measurement of microbial respiration and substrate-induced respiration (SIR)

To quantify this important thing that links to our hypothesis (microbial respiration would be higher in root-containing plots than in root-free plots), we used sieved soils to assess microbial respiration and microbial biomass C in soils with or without plant roots in the laboratory. Following Min et al.⁵⁷, we placed 10 g of fresh soil in 400 mL jars and pre-incubated them at 22 °C overnight (8 jars = 2 treatments * 4 replicates). The jars were sealed with a lid equipped with a non-dispersive infrared CO₂ sensor (K30-FS 10,000 ppm sensor, CO₂ meter). We measured CO₂ concentration every 30 s for 15 min, calculated the slope of the CO₂ flux, and repeated this for 2 h (thus 6 times of measurements of CO₂ flux). The microbial respiration rate was expressed as μg C-CO₂ g⁻¹ dry soil h⁻¹_.

Once microbial respiration rate was assessed, we added 400 mg of powdered yeast extract (Yeast Extract, Sigma-Aldrich, USA) to the jars as a C and nutrient source and adjusted water content to 60% of water holding capacity. The CO₂ concentration in the jar was recorded every 30 s for 15 min, and this measurement was repeated 8 times (substrate-induced respiration, SIR). We then calculated the average SIR in μg C-CO₂ g⁻¹ dry soil h^-1 and converted it to microbial biomass C following Anderson and Domsch⁵⁸.

$$Microbial biomass C \left( {\mu g C g^{ - 1} soil} \right) = \left( {\mu L CO_{2} g^{ - 1} soil h^{ - 1} } \right) \times 40.04 + 0.37$$

(1)

The average microbial biomass C over the entire measurement period served as an index of the SIR-responsive microbial biomass C pool.

Field measurement of R_S and R_H, and the Q ₁₀ of R_S and R_H

We also assessed field respiration from the M. × giganteus plots and the M. × giganteus-free plots (hereafter, R_S and R_H, respectively). In the field, PVC collars (outside diameter: 11.4 cm, height: 7.5 cm) were randomly inserted into the M. × giganteus plots and the M. × giganteus-free plots (n = 4, each) at a depth of 3 cm in early May 2023. All collars were fixed in the same position in the plots during the entire experiment. To eliminate the influence of above-ground plant respiration, small living plants inside the collars were clipped, and the litter within the collar was removed before the measurements of CO₂ fluxes. Soil CO₂ fluxes were measured once or twice every month from May to December in 2023 using the EGM-5 infrared gas analyzer (PP systems, USA) connected to the SRC-2 soil respiration chamber. The CO₂ fluxes in the field data were derived from the slope of the linear increase in the CO₂ concentration over the sampling time, calculated as follows:

$$F_{{CO_{2} }} = \frac{dC}{{dT}}{\text{ x }}\frac{PV}{{ART}}$$

(2)

, where $dC$ is the variation in CO₂ concentration (ppm), $dT$ is the process execution time (s), $P$ is the air pressure (atm), $V$ is the chamber volume (L), $A$ is the area of soil enclosed by the chamber, $R$ is the ideal gas constant (0.082 L atm mol⁻¹ K⁻¹), and $T$ is the soil temperature (K). Soil temperature and soil volumetric water content were measured at a depth of 0–6 cm using Stevens Hydra Probe (PP systems, USA) concomitantly with the probe inserted adjacent to the collar during the soil CO₂ flux measurements. The R_H:R_S ratio was calculated for each measurement day and averaged every month.

To quantify the relationships between soil CO₂ flux and the interaction of soil temperature and soil water content, nonlinear models were fitted using the following equation⁵⁹.

$$FCO_{2} = \left( {\gamma_{1} W + \gamma_{2} W^{2} } \right) \times e^{{r_{3} T}}$$

(3)

, where $F{\text{CO}}_{2}$ measured soil CO₂ fluxes (μg C cm⁻² h⁻¹), $W$ is soil water content (m³ m⁻³), $T$ is the soil temperature (°C), and ${\gamma }_{1}$, ${\gamma }_{2}$, and ${\gamma }_{3}$ are the fitted model coefficients.

An exponential model was used to examine the responses of soil CO₂ fluxes to soil temperature¹:

$$FCO_{2} = \alpha e^{\beta T}$$

(4)

, where $FC{O}_{2}$ measured soil CO₂ fluxes (g C m⁻² yr⁻¹), $T$ the soil temperature (°C), the coefficient α is the soil CO₂ flux at a reference temperature of 0 °C, and β is the sensitivity of soil CO₂ flux to soil temperature. The β value was used to calculate the Q₁₀ value. Note that we calculated Q₁₀ for the non-monsoon period only, as the temperature range during the monsoon period was not wide enough to calculate Q₁₀.

$${\varvec{Q}}_{10} = {\varvec{e}}^{{10{\varvec{\beta}}}}$$

(5)

World and Korea literature search

We retrieved R_S and R_H data from the 5.0 version of the SRDB (SRDB-V5) (downloaded on 30 October 2023 from https://daac.ornl.gov/cgi-bin/dsviewer.pl?ds_id=1827). This comprehensive database collects a broad spectrum of field-measured soil respiration data from 1961 to 2017, incorporating over 2,266 published studies and 10,366 individual records. For soil respiration analysis, we used data that met specific criteria (i) inclusion of both annual R_S and R_H, (ii) R_H:R_S ratios between 0 to 1, (iii) no manipulations such as fertilization or CO₂ enrichment, and (iv) data from forests, grasslands, and agricultural study sites only. The number of data for each ecosystem was as follows: forests (total n = 514; including deciduous forest, n = 202 and evergreen forest, n = 312), grasslands (n = 70), and agricultural lands (n = 48).

Also, we compiled R_S and R_H data from Korean literature using the keywords “soil respiration”, “heterotrophic respiration”, “autotrophic respiration” and “trenching” in the Korean Citation Index, Google Scholar, and Research Information Sharing Service (Supplementary 1). 256 data points were extracted from 9 articles using Web Plot Digitizer (https://automeris.io/WebPlotDigitizer/). These data were compiled with specific criteria similar to SRDB, covering forests (n = 255), grassland (n = 1), and agriculture (n = 0).

Statistical analyses

We conducted the Shapiro–Wilk test to check if data are normally distributed. The initial biological and physicochemical properties in the M. × giganteus and M. × giganteus-free soil samples were compared by using independent sample t-tests. We used the Wilcoxon rank-sum test to compare root biomass between the M. × giganteus and M. × giganteus-free soils. We used Spearman correlation to determine the relationship between microbial biomass C and soil CO₂ flux, and Pearson correlation for microbial biomass C and K₂SO₄-extractable OC. Additionally, we used one-way analysis of variance (ANOVA) to compare the SRDB and Korean data, and to examine the differences in the R_S and R_H across different ecosystems. Two-way ANOVA was used to assess the effects of root exclusion and time on microbial respiration, soil CO₂ flux, soil temperature, and soil water content. All statistical analyses were performed using R 4.3.1, with statistical significance set at p = 0.05⁶⁰.