Soil warming increases soil temperature sensitivity in subtropical Forests of SW China

Site description

The experimental field is located at site of the Ailaoshan Station for Subtropical Forest Ecosystem Studies (24°32′N, 101°01′E; 2,480 m above sea level) of the Chinese Ecological Research Network. According to monitoring data for 2002 to 2011, the annual average air temperature was 11.3 °C; the maximum air temperature was 15.6 °C (July); the minimum temperature was 5.7 °C (January); the annual average precipitation level was 1,778 mm, and rainfall levels reached close to 85% of total annual rainfall for the rainy season (May–October) (Wu et al., 2014). The main type of soil found in the area yellowish-brown with a pH value of 4.5; soil organic carbon levels reach 303.8 g kg⁻¹ and total nitrogen levels reach 18.38 g kg⁻¹ in the humus horizon (Chan et al., 2006). The main species found in the area include Lithocarpus hancei, Castanopsis wattii, Schima noronhae, and Lithocarpus xylocarpus (Wu et al., 2014).

Experimental design

In the subtropical forest, six plots of 1 m × 1 m were randomly selected, and two treatments were established, including control and warming treatments with three replicates. A carbonizing infrared radiator was applied as the heat radiation source for the warming treatment, and a 200-watt carbon infrared radiant heating lamp (each lamp was 90 cm long and each lampshade was 100 cm long, 15 cm wide, and 10 cm deep) was placed approximately 1.0 m above the sample side of each plot. A continuous heating mode was adopted from June 2016 to May 2017, supplying continuous power except during power outages. In each plot, a PVC pipe (inner diameter of 20 cm and soil depth of two cm) was placed to directly couple with an external PVC pipe (outer diameter of 20 cm, upper seal base opening and height of 20 cm) for further measurements of R_S.

Data collection and calculations

Soil CO₂ efflux was measured monthly using a gas analyser (Li-820; Li-Cor, Lincoln, NE, USA) between 9:00 and 11:00 AM (Beijing Time) to measure R_S levels. Soil temperatures (T, °C) were measured with a digital thermometer (6310; Spectrum, Illinois, USA), and soil water content (W, %) was measured via time domain reflectometry (MP-KIT; Beijing Channel, Beijing, China) at a depth of 50 mm. The R_S rate was calculated as follows:

(1)${\displaystyle R_S=\frac{V\cdot P}{R\cdot S\cdot Ta}\cdot \frac{dc}{dt}}$

where R_S is the soil CO₂ efflux value (µmol m⁻² s⁻¹); V is the chamber volume (m³); S is the chamber area (m²); R is the gas constant (8.314 Pa m³ k⁻¹ mol⁻¹); T a is air temperature (K); P is air pressure (Pa), and dc/dt is the slope of variations in CO₂ concentrations overtime for the measurement period.

We used a one-factor regression to determine the relationship between R_S and T and a two-factor regression to determine relationships of R_S with T and W. The regression equations applied are as follows (Xu & Qi, 2001a; Liu et al., 2018):

(2)${\displaystyle R_S=a\cdot e^{bT}}$ (3)${\displaystyle R_S=a\cdot e^{bT}\cdot W^c.}$

Where a, b and c are estimated from the regressions; b is the soil Q₁₀ coefficient, and c is the soil water sensitivity coefficient.

Q₁₀ is calculated using the following equation (Luo et al., 2001):

(4)${\displaystyle Q_{10}=e^{10b}.}$

Where b is derived from Eqs. (2) and (3).

Different Q₁₀ values derived from the one- and two-factor regressions are attributable to soil water effects, and the effect on Q₁₀ (WE_Q10) was calculated with the following equation:

(5)${\displaystyle W{E}_{Q10}=\frac{Q_{10-\mathrm{one}-\text{factor}}-Q_{10-\mathrm{two}-\text{factor}}}{Q_{10-\mathrm{two}-\text{factor}}}\times 100\text{\%}}$

The parameters were estimated from the models listed above using the nonlinear regression dynamic fit wizard, and t tests were used to test differences in the R_S, T, W, and Q₁₀ values of the control and warming treatments (Version 12; Systat Software, Inc., Point Richmond, California, USA).

Effect of warming on soil environment factors

The warming treatment increased the soil temperature but did not change seasonal variation patterns (Fig. 1A). Mean annual soil temperatures measured from the control and warming treatments were recorded as 11.1 ± 0.1 °C and 12.5 ± 0.1 °C, respectively, and warming significantly increased the soil temperature by 1.4 °C (Fig. 1B). During the measurement period, the warming treatment significantly affected soil temperatures during both the dry and rainy seasons (Fig. 1B). However, warming did not change soil water content levels, and in terms of seasonal variations, we found no significant differences between dry season, rainy season and yearly trends (Figs. 1C & 1D). Mean annual soil water levels measured from the control and warming treatments were recorded as 30.3 ± 3.0% (v/v) and 29.0 ± 2.8% (v/v), respectively, reduced by 4.2% (%/%) (Fig. 1D).

Effect of warming on soil respiration

Soil warming increased R_S levels in the rainy season but suppressed them in the dry season (Fig. 2A). No significant differences in means of the control and warming treatments were observed across dry season, rainy season, and annual period (Fig. 2B). In the dry season, mean values of the two treatments were recorded as 2.68 ± 0.47 µmol m⁻²s⁻¹ and 2.27 ± 0.28 µmol m⁻²s⁻¹, respectively, with the warming treatments showing a reduction of 15.0%. In the rainy season, mean values for the two treatments were recorded as 5.50 ± 0.79 µmol m⁻²s⁻¹ and 6.50 ± 0.32 µmol m⁻²s⁻¹, respectively, with the warming treatments showing an increase of 18.1%. The mean annual R_S for the two treatments was recorded as 4.09 ± 0.28 µmol m⁻²s⁻¹ and 4.39 ± 0.25 µmol m⁻²s⁻¹ where values increased by 7.3% during the warming treatment.

Effect of warming on temperature sensitivity

From one-factor regression model results, the Q₁₀ values of the control and warming treatments were recorded as 2.6 ± 0.7 and 4.9 ± 0.3, respectively, and warming significantly increased the Q₁₀ value by 89.1%. According to two-factor regression model results, the Q₁₀ values reached 2.0 ± 0.5 and 3.3 ± 0.2, respectively, and warming significantly increased the Q₁₀ values by 67.4% (Fig. 3). One- and two-factor regression parameters are given in Table 1. WEQ₁₀ increased when the c values fell below 0.779 and decreased when the c values exceeded 0.779 (Fig. 4).