Global soil CO2 efflux was estimated to be 80.4 Pg C yr−1 from 1980 to 19941, a value that had increased to 98.0 Pg C yr−1 by 20082. Global soil organic carbon was estimated to be about 3300 Pg, four times the amount present in living plants globally3,4. There is a positive relationship between global soil CO2 efflux and temperature with an increase of 3.3 Pg C yr−1 per °C1, indicating that increased temperature has a great potential to affect soil CO2 efflux and produce a positive carbon–climate feedback under global warming. Forests cover only about 30% of the terrestrial ecosystems, but contain about 45% of terrestrial carbon5. Therefore, forest soil organic carbon decomposition plays an important role in the regional and global carbon cycle and studies have increasingly been conducted to explore the effects of warming on forest soil organic carbon decomposition, aiming to improve prediction of carbon–climate feedbacks.

Many previous studies focused on the warming effects on soil respiration6,7,8,9,10. However, soil respiration includes two main components: heterotrophic respiration (RH) and autotrophic respiration (RA). Responses of heterotrophic respiration (RH) and autotrophic respiration (RA) to warming are not always the same. For example, Schindlbacher, et al.11 reported that RH and RA responded similarly to temperature increase; both increasing with warming. However, Zhou, et al.12 reported a negative effect of warming on both RH and RA. Other studies showed that warming increased RH, but decreased RA13,14. Because only the RH component of soil respiration has potential to increase atmospheric CO215, it is important to separate RH from soil respiration and to examine how warming may affect RH16,17,18.

Forest soil organic carbon decomposition plays an important role in the carbon cycle, but most previous studies about WERH were carried out in grassland ecosystems12,13,14,19,20,21. A few studies have been conducted in boreal coniferous forest dominated by Norway spruce11,22,23 and in a temperate forest18. To our knowledge, no studies have been reported in subtropical forests.

The subtropical evergreen broad-leaved forest in the Ailao Mountains in Yunnan of southwestern China has been considered a carbon sink24, but the carbon sink strength is likely to be weakened by warming25. During the period from 1961 to 2004, the mean annual air temperature over the Yunnan Plateau (southwestern China) increased at a rate of 0.3 °C per decade26. A recent study showed that the mean annual air temperature increased by 0.36 °C per decade from 1983 to 2010, leading to an increasing trend of 0.31 °C per decade from 1986 to 2010 in the top 10 cm of soil in the subtropical forest27. Previous studies have shown that RH exhibited seasonal variation and had a significant positive relationship with soil temperature28,29. However, it remains unclear how soil warming affects RH in this subtropical forest.

Previous studies have showed inconsistent results about WERH. A few studies suggested that warming decreased RH12,20, while some demonstrated a positive effect on RH which can be sustained for 5–10 years18,30,31. Other studies argued acclimatization in terms of the warming effect declining over time8,22,32, which may be ascribed to depletion of soil labile carbon16,19,33,34 or acclimatization of soil microbes35,36,37,38. Other studies also argued acclimatization in terms of temperature sensitivity adaption11,17,21. Based on mentioned studies above, we hypothesized that RH acclimated to continuous warming in this forest. To test this hypothesis, we conducted ‘trenching alone’ and ‘trenching with warming’ treatments in this forest.


Treatment effects on soil environmental factors and soil carbon efflux

Compared with the control treatment (CK), the trenching treatment (NR) did not change soil temperature (T, °C) (Fig. 1a), but increased soil water content (W, % (v/v)) (Fig. 1b). The trenching together with warming treatment (NRW) increased T (Fig. 1a), but did not change W, except for the first 1.5 years (Fig. 1b). Soil carbon efflux reduced in the NR treatment in the first year and changed little in the later years during the measurement period. In the NRW treatment, soil carbon efflux changed little (Fig. 1c).

Figure 1
figure 1

Seasonal variations of soil temperature (a), soil water content (b) and soil carbon efflux (c) after treatments began. Black lines with grey shadows represent the control treatment (CK), cyan lines with semi-transparent cyan shadows represent the trenching treatment (NR) and red lines with semi-transparent red shadows represent the trenching together with warming treatment (NRW) (Mean ± SE).

Compared with the NR treatment, the NRW treatment increased T, but decreased W (by annual average of 5.1%, v/v) (Fig. 1a,b) and increased RH especially in the first year (Fig. 1c).

There were similar seasonal variations in R, T and W of the CK, NR and NRW treatments (Fig. 1). The Pearson correlation analysis showed R more correlated to T (Table S1) and the two-factor regression model showed that R had a positive relationship with both T and W (Fig. 2).

Figure 2
figure 2

Two-factor regression of soil carbon efflux with soil temperature and soil water content.

Warming effects on RH and T

To clarify the effect of warming on RH, we focus on the affecting factor of changes to soil water content. Two-factor regression models (Fig. 2) were used for correction (for details see the data analysis section).

Results show that RH showed seasonal variation in both the NR and NRW treatments, with maximum values appearing in July or August and minimum values occurring in February. The difference value between NR and NRW (RNRWRNR) also exhibited seasonal variation (Fig. 3a). WERH had similar seasonal variation to RH with about 3 months delay; during the measurement period, WERH was 33.9, 23.8, 18.5 and 15.3% for 2011–2014, respectively (with an average of 22.9%), clearly decreasing year by year (Fig. 3b). Variation of the warming effect on T (WET) was very similar to that of WERH and WET was 2.6, 2.1, 2.0 and 1.8 °C (with an average of 2.1 °C) for 2011–2014, respectively (Fig. 3c).

Figure 3
figure 3

Variation of heterotrophic respiration in warmed (NRW) and unwarmed (NR) treatments and their difference (RNRWRNR) (a), the variation of the warming effect on heterotrophic respiration (WERH) (b) and the variation of the warming effect on soil temperature (WET) (c).

Factors affecting the warming effect on RH

The factors T, W and WET had a significant relationship with WERH (p < 0.0001), which meant WERH was affected by T, W and WET. The factors T, W and WET explained 13, 33 and 75% of WERH’s variation, respectively. A 1 °C increase in T and a 1% decrease in W would reduce 0.98 and 0.79% of WERH, respectively; however, a 1 °C increase in WET would cause a 21.78% increase in WERH (Fig. 4). This suggests that a range of increased soil temperature (WET) control WERH.

Figure 4
figure 4

Relationships between WERH and T (a), W (b) and WET (c).


Previous studies have shown that RH acclimated to warmer temperature over the warming period8,22,32. However, several long-term warming experiments have shown sustained WERH in ecosystems with high soil organic carbon18,30,31. This indicates that such sustained stimulation may be attributed to high substrate availability18. In this study, soil organic carbon, soil labile organic carbon and soil microbial carbon did not significantly (p > 0.05) decrease in NR and NRW comparing to CK during the 4-year measurement period, but they were larger in NRW than that in NR (Fig. S2). In this case, during the 4-year measurement period, substrate is unlikely to be a limiting factor28,39. Our results show that WERH declined year by year. Without further analyses, this indicated “acclimatization” superficially (Fig. 3b). However, this decline probably was not caused by acclimation but was caused by the reduction in WET over the 4-year observation period (Figs 3c and 4c). The temperature sensitivity (Q10) analysis showed that warming increased Q10 in the later period of this study which showed opposite result to previous studies11,17,21 and indicated the absence of “acclimatization” (Fig. S3). A current meta-analysis showed that WERH remained stable during warming period, which also challenged “acclimatization” of microbial activity to warmer temperature40. Therefore, WET is likely to be the controlling factor in this study.

In this study, WERH (22.9%) was lower than that of Aguilos, et al.18, Schindlbacher, et al.11 and Eliasson, et al.22 reported (82.0%, 42.0% and >30%, respectively). WET values in the experiments of Aguilos, et al.18, Schindlbacher, et al.11 and Eliasson, et al.22 were 3, 4 and 5 °C, respectively, all larger than ours (2.1 °C). A review showed that warming by 2.0 °C increased RH by an average of 21.0%40, similar to our result. Another 2-year soil monolith transplantation experiment at the same site demonstrated that soil organic carbon efflux increased by 70.5 and 62.6% because of an annual WET of 3.9 and 6.7 °C (unpublished data, Fig. S4). These results suggest that a positive relationship between WERH and WET requires a specific range of WET and when WET exceeds this range, WERH becomes restricted (Fig. S4). Therefore, WET controlled WERH in this subtropical forest.

Although soil temperature is more important than soil water content on affecting dynamics of soil carbon efflux (Fig. 1, Table S1), dynamics of soil water content may also affect WERH. For example, Liu, et al.20 found that warming decreased soil water content, which resulted in a reduction in RH. Similarly, a prolonged summer drought offsets soil warming effects9. Our results also suggest that soil water content affected WERH, mediating WERH under the warmer soil temperature conditions (Fig. 4b).

In this study, WET decreased yearly, perhaps owing to aging of warming lamps. WET also showed seasonal variation (Fig. 3c). These factors should be considered in future research when researchers are attempting to maintain a constant WET. Nevertheless, these results allowed us to analyse the WET effect on WERH. In this forest, RH is 9.53 t C ha−1 yr−1 (Fig. 3a) and given an increase of 1 °C in WET, WERH will increase by 21.78%, an additional soil carbon release to the atmosphere can be estimated to be about 2.65 t ha−1 yr−1 from the subtropical forest.

Response of soil organic carbon to temperature depends on soil temperature41 and quality of soil organic carbon42,43. Factors affecting soil carbon quality and decomposition processes can influence the temperature response and thus warming effect44. As soil water also affects soil carbon efflux, it’s changes should be also considered to prediction of carbon–climate feedbacks40,45. Further studies should focus on responses of WERH and interactions between factors.


In summary, our warming experiment increased soil temperature by 2.1 °C and thereby increased heterotrophic respiration by 22.9% during the 4-year observation period. Continuous measurement allowed us to analyse variations of T, W, WET and WERH and their inter-relationships. Our results show that the warming effect on RH was controlled by a range of increased soil temperature and regulated by the variation of soil water content. This suggests that global warming will accelerate soil carbon efflux to the atmosphere, regulated by a change of soil water content in response to warming and rainfall changes. Future warming in subtropical forests can accelerate release of soil organic carbon to the atmosphere.

Materials and Methods

Site description

This experiment was conducted at the Ailaoshan Station for Subtropical Forest Ecosystem Studies (24°32′N, 101°01′E; 2480 m above sea level) of the Chinese Ecological Research Network, which is located in Jingdong County, Yunnan Province, China. The annual mean air temperature was 11.3 °C, with a minimum monthly mean temperature of 5.7 °C in January and a maximum monthly mean temperature of 15.6 °C in July. Average annual rainfall was 1778 mm, with 86.0% of this falling in the rainy season (May–October)28. The dominant tree species in the forest are Lithocarpus xylocarpus, Lithocarpus hancei and Castanopsis wattii. The soils are Alfisols with a pH of 4.5, soil organic carbon of 304 g kg−1 and total nitrogen of 18 g kg−1 in the humus horizon39.

Data collection

A multichannel automated chamber system for continuous measurement of soil CO2 effluxes was established in October 2010. The system comprised 20 automatic chambers (90 × 90 × 50 cm) and a control box (Fig. S1). On 17 December 2010, the 20 chambers were divided into four treatments (five chambers per treatment): control (CK), litter removal (NL), trenching (NR) and infrared light warming together with trenching (NRW). In this study, we only discuss the NR and NRW treatments and the control CK. The infrared light warming method has been frequently applied for soil warming in forest ecosystems3,34,46,47 and is also used in this study. The key advantages of this method are that there is no disturbance to the soil structure and it involves the same process of warming as heating from the climatic warming effect (heating soil from the surface to a deeper depth). For trenching treatments, a 1 m × 1 m square trench (width 30 cm, depth 50 cm) was dug to form a cube of soil contained by PVC planks; soil was backfilled by its original layers with topsoil over subsoil. The main components of the control box are an infrared gas analyser (Li-820, Li-Cor Inc., Lincoln, NE, USA) and a datalogger (CR1000, Campbell Scientific Inc., Logan, UT, USA), as described in more detail in earlier studies48,49.

Soil temperature (T, °C) at a 5-cm depth and air temperature (Ta, °C) were measured inside each chamber using self-made thermocouples49. Soil water content (W, % (v/v)) at a 10-cm depth was monitored using time domain reflectometry (CS-616, Campbell Scientific Inc.)49. Air pressure (P, hPa) at a 30-cm height in the centre of the plot was measured by a pressure transducer (PX2760, Omega Engineering, Inc., Stamford, CT, USA)49.

Soil C–CO2 efflux (R) was calculated as follows:

where R is soil C–CO2 efflux (g C–CO2 m−2 d−1); M is the C molar mass; V0, P0 and T0 are constants (22.4 L mol−1, 1013.25 hPa and 273.15 K, respectively); Ta and P are air temperature (K) and pressure (hPa), respectively; H the is height of the respiration chamber (m); and dc/dt is the slope of CO2 concentration variation with time over the measurement period (measurement for 3 min and the calculation ignores data from the first minute).

Data analysis

Every day, each chamber received 24 data points for R (hourly) and 48 data points for T and W (twice per hour), except during periods of electrical failure. For each chamber, available data were used to calculate daily average values. Previous analysis showed that RH was unusual in the fifth group for both NR and NRW (Fig. S1); therefore, the fifth group was ignored, leaving four groups as the four repeats in this study.

The two-factor regression model was used to quantify the relationship of soil temperature (T) and soil water content (W) with soil carbon efflux (R) as follows50:

where a, b and c are constants estimated from the regression model by the nonlinear regression dynamic fit wizard using Sigmaplot (Version 12.5, Systat. Software, Inc., Point Richmond, CA, USA) (for details see Fig. 2).

Figure 2 suggests that T and W had a positive effect on R, while treatments changed the soil microclimate (Fig. 1a,b). Compared to CK, NR increased soil water content; however compared to NR, NRW decreased soil water content (Fig. 1b). To eliminate the biases due to changed soil water content, WERH should be calculated under the same soil water content condition. Therefore, the background soil temperature and soil water content measured in the control treatment were used to correct the values of RNR, using Equation (2). For NRW, the soil temperature of the NRW plots and the background soil water content of the control plots were used for the correction, so that the warming effect was due to the soil temperature increase without the effect of soil water content decrease (Fig. 3b). Parameters (a, b and c) for NR and NRW were shown in Fig. 2b,c and their estimation results were shown in Table S2.

Warming effects on RH (WERH, %) and on soil temperature (WET, °C) were calculated using the following equations:

Additional Information

How to cite this article: Wu, C. et al. Heterotrophic respiration does not acclimate to continuous warming in a subtropical forest. Sci. Rep. 6, 21561; doi: 10.1038/srep21561 (2016).