Prediction of annual soil respiration from its flux at mean annual temperature
Introduction
Soil respiration (SR, 68-98 Pg C yr−1), is the largest carbon flux from the terrestrial surface to the atmosphere (Bond-Lamberty and Thomson, 2010). Robustly scaling SR across time and space is important to understand how terrestrial carbon responds to global climate change. The accuracy of the spatial and temporal scaling of SR depends on its measurement frequency (Gomez-Casanovas et al., 2013; Jian, et al., 2018), but SR is difficult to measure continuously. Older methods such as alkali trapping require long absorption times and it may underestimate the flux (Rochette and Hutchinson, 2005), while static chambers require taking gas samples in a fixed interval over a certain period (e.g., 15–30 mins), with subsequent analysis in a gas chromatograph, making it difficult to measure frequently or at night (Giasson et al., 2013). Long-term chamber systems with integrated infrared gas analyzers can measure SR continuously, but the equipment is expensive, and difficult to maintain for long time periods (Luo and Zhou, 2006). In remote or harsh or seasonally snow-covered regions such as the Arctic, it is thus very hard to accurately estimate the annual SR flux, even though year-round measurements are particularly needed in these regions (Elberling, 2007; Xu and Shang, 2016).
For these reasons, it is common to use SR from a short period in time to represent the mean SR over a (potentially much longer) period. For example, SR is usually not measured at night (Hu et al., 2016) but rather a single daytime measurement is assumed to represent the average of a 24-hour day (SRdiurnal); SR measured once or twice a month to represent monthly mean RS; and 4-12 months of data used to estimate an annual flux (SRannual) (Jian et al., 2018a). Theoretically, if SR increased linearly with temperature (i.e., Q10 = 1), and no other limiting factors were involved, SR measured at some mean soil temperature (SRMAST) might indeed equal its long-term average (Bahn et al. 2010; Jensen et al., 1996). However, SR does not change linearly as soil temperature (ST) increases (i.e., Q10 ≠ 1), but usually follows an exponential or second order exponential trend (Jensen et al., 1996; Lloyd and Taylor, 1994). For these reasons–Jessen's inequality, the non-linear relationship between SR and temperature, and other factors operating at longer timescales–SR at a single timepoint usually cannot be used to directly estimate the annual flux SRannual (Ruel and Ayres, 1999; Smallwood, 1996).
Addressing this problem, Bahn et al. (2010) found that SR measured at the mean annual soil temperature (SRMAST) was highly correlated with SRannual across 80 site-year records worldwide, and they developed a model to predict SRannual from SRMAST. This insight potentially allows for easily estimating SRannual under ‘normal’ conditions, and this approach has been evaluated at the site scale (e.g. Oishi et al., 2013). However, the robustness of using SRMAST to estimate SRannual across different biomes, ecosystems, and precipitation conditions should be evaluated by more independent data globally to account for the lager variation of background context (e.g., climate, soil, and vegetation types). Furthermore, Bahn et al. (2010) used the relationship between SR and soil temperature (ST) to calculate SRMAST, which was then compared to SRannual. Many (but not all) studies used the same SR~ST relationship, however, to calculate the ‘true’ SRannual flux, meaning these are not truly independent data; it is thus difficult to determine whether the tight relationship between SRMAST and SRannual is due to autocorrelation or to a true relationship between them (Giasson et al., 2013). As a result, using independent SRMAST (e.g., SR measured within 1 ℃ rather than derived from a Q10 equation) from multiple sites across the globe will provide additional evidence as to whether SRMAST can robustly predict SRannual.
In addition, it is important to test whether SR at mean annual air temperature (MAAT) can predict SRannual. Bahn et al. (2010) showed that SRMAST at mean ST highly correlated with SRannual, but the lack of (and uncertainty surrounding) soil temperature data, from site-specific (Tabari et al., 2015) to macro (Bell et al., 2013; Hu et al., 2002) scales, precludes its use to predict SRannual across large spatial scales. That is, the Bahn et al. (2010) model depends on data typically measured on-site, not remotely, limiting its applicability. Conversely, air temperature is accurately available globally, opening the door to more widespread and accurate predictions.
Here we use comprehensive global SR databases (from 387 different sites and comprising 823 site-years of data) to investigate whether SR from mean annual soil temperature (MAST) or mean annual air temperature (MAAT) can predict SRannual robustly, subjecting the Bahn et al. (2010) model to stringent tests for bias and error. Specifically, we aimed to (1) explore different ways to estimate SRMAST and assess the strengths, bias, and errors of each; (2) evaluate whether SR at MAAT can predict SRannual; and 3) investigate the robustness of using SRMAST or SRMAAT to predict SRannual across diverse ecosystem types and climatic conditions.
Section snippets
Data collection and processing
To test the robustness of using SRMAST for estimating mean SR over both short (day) and long (year) timescales, we used data from a global diurnal soil respiration dataset (Jian et al., 2018a) and data from a recent version of the global soil respiration database (SRDB_V4, downloaded from https://github.com/bpbond/srdb; also available at the Oak Ridge National Laboratory DAAC, https://daac.ornl.gov/cgi-bin/dsviewer.pl?ds_id=1578). We limited the SR records to those from studies that were
Predicting mean SR of a period from SR at mean soil temperature
SR at mean temperature was highly correlated with mean SR rate at both diurnal and annual time scales (Fig. 2 and Table 1). At the diurnal timescale, the slope of the linear model between SR at daily mean ST and SRdiurnal was 0.94 (±0.02); the intercept was 0.12 g C m−2 day−1, close to 0; Diurnal SR at mean ST explained 93% of SRdiurnal variability (Table 1). As expected, at the annual timescale, the divergence between SRMAST and mean SRannual was larger: the slope of the linear model between SR
Discussion
We found that both SRMAST and SRMAAT can be used to predict SRannual, with well-quantified errors. The divergence between SRannual and SRMAST or SRMAAT, however, increased as the timescale lengthened (i.e., the divergence was larger at the annual than daily timescale, Fig. 2). This is unsurprising; environmental conditions (e.g., soil moisture, soil carbon content, microbial community, microbial activity, and vegetation conditions) usually do not drastically change within a day, and therefore
Conclusion
Studies have two important problems for SR and more generally carbon-cycle measurement and modeling: we have many more SR measurements in mid-latitude regions and developed countries. Less-developed countries are constrained by lack of resources, other regions are constrained by the extreme weather during winter, and thus we do not have enough measurements from southern hemisphere, arctic, and tropical regions (Xu and Shang, 2016). It is difficult to measure soil respiration all year around in
Declaration of Competing Interest
None.
Acknowledgement
This research was supported by the DOE Office of Biological and Environmental Research (BER), as part of BER's Terrestrial Ecosystem Science Program [grand number: DE-AC05-76RL01830]. All code and data to reproduce all results in this study are available at https://github.com/PNNL-TES/bahn-rs-test.
References (45)
- et al.
Influence of urban land development and soil rehabilitation on soil–atmosphere greenhouse gas fluxes
Geoderma
(2014) - et al.
Carbon budget study using CO2flux measurements from a no till system in central Ohio
Soil Tillage Res.
(2000) Annual soil CO2 effluxes in the High Arctic: The role of snow thickness and vegetation type
Soil Biol. Biochem.
(2007)- et al.
Alpine grassland ecosystem respiration variation under irrigation in Northern Tibet
Acta Ecol. Sin.
(2014) - et al.
Short-term response of soil respiration to nitrogen fertilization in a subtropical evergreen forest
Soil Biol. Biochem.
(2014) - et al.
Soil surface CO2 flux as an index of soil respiration in situ: A comparison of two chamber methods
Soil Biol. Biochem.
(1996) - et al.
Measurement strategies to account for soil respiration temporal heterogeneity across diverse regions
Soil Biol. Biochem.
(2018) - et al.
Comprehensive comparison of gap-filling techniques for eddy covariance net carbon fluxes
Agric. For. Meteorol.
(2007) - et al.
Spatial and temporal variability of soil CO2 efflux in three proximate temperate forest ecosystems
Agric. For. Meteorol.
(2013) - et al.
Jensen's inequality predicts effects of environmental variation
Tree
(1999)
Factors controlling Eucalyptus productivity: How water availability and stand structure alter production and carbon allocation
For. Ecol. Manage.
Carbon neutral or a sink? Uncertainty caused by gap-filling long-term flux measurements for an old-growth boreal black spruce forest.
Agric. For. Meteorol
Simulated rain addition modifies diurnal patterns and temperature sensitivities of autotrophic and heterotrophic soil respiration in an arid desert ecosystem
Soil Biol. Biochem.
On the “temperature sensitivity” of soil respiration: Can we use the immeasurable to predict the unknown?
Soil Biol. Biochem
The effect of nitrogen addition on soil respiration from a nitrogen-limited forest soil
Agric. For. Meteorol.
Contribution of soil respiration to the global carbon equation
J. Plant Physiol.
Soil respiration at mean annual temperature predicts annual total across vegetation types and biomes
Biogeosciences
U.S. climate reference network soil moisture and temperature observations
J. Hydrometeorol.
Temperature-associated increases in the global soil respiration record
Nature
Simulated chronic NO-3deposition reduces soil respiration in northern hardwood forests
Glob. Chang. Biol.
Analysis of the spatial and temporal changes in soil CO2 flux in alpine meadow of Qilian Mountain
Environ. Geol.
Soil water content and temperature as independent or confounded factors controlling soil respiration in a temperate mixed hardwood forest
Glob. Chang. Biol.
Cited by (16)
BESSv2.0: A satellite-based and coupled-process model for quantifying long-term global land–atmosphere fluxes
2023, Remote Sensing of EnvironmentEffect of tillage and crop type on soil respiration in a long-term field experiment on chernozem soil under temperate climate
2022, Soil and Tillage ResearchCitation Excerpt :Rs is a highly uncertain part of the global carbon cycle, potentially affecting atmospheric CO2 concentrations under changing climatic conditions (Carey et al., 2016; Hursh et al., 2017; Jones et al., 2003). The primary abiotic driver of microbial activity is soil temperature (Ts) (Lloyd and Taylor, 1994), and it has recently been shown that Rs measured at annual mean temperature is a good predictor of annual mean Rs (Jian et al., 2020). Under certain conditions, soil water content (SWC) also affects Rs (Balogh et al., 2016; Orchard and Cook, 1983; Reichstein et al., 2003).
Response of soil respiration and Q<inf>10</inf> to temperature and moisture in naturally regenerated and bare lands based on an 11-year observation period
2022, CatenaCitation Excerpt :This also implied that in semiarid areas, the annual Rs could be sensitive to precipitation patterns, especially in the summer drought season and grassland ecosystems (Liu et al., 2009) because the abrupt increase in θ under dry conditions could greatly accelerate Rs (Li et al., 2008; Zheng et al., 2021). The negative relationship between the annual Rs and the annual Ts in this study (Fig. 7b) did not fit the general conclusion that the annual Rs was increased with an increasing annual Ts on a global scale (or on an annual scale in the current study) (Jian et al., 2020; Tang et al., 2020) owing mostly to the large temperature span, which may have masked the effect of θ on Rs. As other studies have reported, the interannual variations in Rs were controlled primarily by θ, and the response of Rs to an increased Ts was constrained by a low θ (Asensio et al., 2007) or annual precipitation (Zhao et al., 2019).
Rates of soil respiration components in response to inorganic and organic fertilizers in an intensively-managed Moso bamboo forest
2021, GeodermaCitation Excerpt :Thus, the soil RA would be less dependent on the amount of labile organic C fraction in soils. In intensively-managed forests, the practices used can markedly alter the soil RH rate by changing the activities of soil enzymes (Jian et al., 2020; Lull et al., 2020). For example, Tu et al. (2011) revealed that an increase in the RH rate from a Pleioblastus amarus forest soil was strongly correlated with an increased activity of soil invertase in the treatment of NH4NO3 fertilizer.
Global consistency in response of terrestrial ecosystem respiration to temperature
2021, Agricultural and Forest MeteorologyCitation Excerpt :Hence, when attempting to represent Re at the global scale, although modellers could use the sigmoid function universally, all biomes with their intrinsic and specific response of Re to temperature must be properly represented with different sigmoid function parameters. Previous studies pointed out that respiration at MAT can be a proxy of annual total soil respiration (Bahn et al., 2010; Jian et al., 2020). Similarly, we found that RN has an excellent match with Rex at MAT, and it is also well represented by the value on the sigmoid curve corresponding with MAT (Fig. 6), both with higher R2 (>0.9).