Interdependencies between temperature and moisture sensitivities of CO 2 emissions in European land ecosystems

Introduction Conclusions References

laboratory incubation experiment.
Emission measurements of carbon dioxide under controlled conditions were conducted using soil monoliths of nine sites from the ÉCLAIRE flux network. Sites are located all over Europe; from the UK in the west to the Ukraine in the east; Italy in the south to Finland in the north and can be separated according to four land-uses 10 (forests, grasslands, arable lands and one peatland). Intact soil cores were incubated in the laboratory at the temperatures 5, 10, 15, 20, and 25 • C in a two factorial design of five soil moisture levels (5, 20, 40, 60, 80 (100) % water filled pore space, WFPS), before analysed for CO 2 fluxes with an automated laboratory incubation measurement system. Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | other factors are not limiting (Ferréa et al., 2012;Meixner, 2006). A widely used term to describe the temperature sensitivity of SOM decomposition is the Q 10 value which is calculated as the proportional increase in CO 2 efflux for a 10 • C increase in temperature . In the context of this paper, we use the term "temperature sensitivity of SOM decomposition" to refer to the short-term temperature dependence 5 of organic matter decomposition as described in Kirschbaum (2006). Other authors reported that land use/cover types, soil moisture content, quality of SOM and temperature itself were found to affect the Q 10 value of soil CO 2 efflux (Shrestha et al., 2004;Wang and Fang, 2009). Temperature sensitivity of SOM decomposition increases with decreasing SOM lability and therefor increasing recalcitrance of SOM (Conant et al., 10 2008;Lützow and Kögel-Knabner, 2009; Thornley and Cannell, 2001;Zimmermann and Bird, 2012) due to the higher activation energy associated with the breakdown of recalcitrant substrates that result in a greater temperature sensitivity of decomposition (Davidson and Janssens, 2006;Hartley and Ineson, 2008). The Arrhenius equation predicts that the Q 10 of chemical reactions decreases with 15 increasing temperature, as is also commonly observed in nature (Kirschbaum, 1995). The theoretical explanation for this negative correlation is that as temperature increases, there is a declining relative increase in the fraction of molecules with sufficient energy to react (Ågren and Wetterstedt, 2007;Davidson and Janssens, 2006). Tuomi et al. (2008) could show that the relationship between temperature and heterotrophic 20 soil respiration can be described best using a Gaussian model. The effect of soil moisture is more complex. Soil water influences the rate of O 2 supply and thereby determines whether aerobic or anaerobic processes prevail within the soil (Pilegaard et al., 2006;Schindlbacher, 2004). The water content is also important for the substrate supply for soil microorganisms (Meixner, 2006). Highest CO 2 25 emissions have been reported at intermediate moisture content while at dry and wet conditions CO 2 emissions decline (Schaufler et al., 2010;Suseela et al., 2012). However, if soil moisture becomes limiting, CO 2 fluxes are suppressed irrespective of high soil temperatures (Davidson et al., 1998;Garten et al., 2008).

4436
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | To describe the effect of moisture on soil microbial activity quadratic functions are common (Moyano et al., 2013;Rodrigo et al., 1997). Moyano et al. (2012) calculated moisture sensitivity as the proportional response of soil microbial respiration to a 0.01 increase in soil moisture of a certain unit. Moisture sensitivity showed highest values at dry conditions decreasing progressively with increasing moisture content. 5 Janssens and Pilegaard (2003) and Qi et al. (2002) expect a positive relationship between temperature sensitivity and moisture content due to the assumption that the effects of soil temperature and moisture are negatively correlated. Thus, soil moisture would be positively correlated with the temperature sensitivity of soil respiration. While Mäkiranta et al. (2009) andGaumont-Guay et al. (2006) found an actual posi-10 tive relationship between temperature sensitivity and soil moisture content in their field measurements Peng et al. (2009) describes in a review of 52 papers (all field measurements) a negative correlation between Q 10 values and mean annual precipitation. However, Curiel Yuste et al. (2004) presented a case study of how the seasonal Q 10 of soil respiration calculated from field measurements can be decoupled from the tem-15 perature sensitivity of heterotrophic soil respiration indicating that the large differences in seasonal Q 10 do not represent differences in the temperature sensitivity of the soil microbial metabolism.
Land use influences the production and consumption of soil CO 2 emissions through vegetation type (Raich andTufekciogul, 2000), root density, N input (Skiba et al., 1998) 20 and management (Flechard et al., 2005). Peng et al. (2009) even found differences of Q 10 values between ecosystem types by comparing field measurements.
In field studies the seasonal development of soil temperature and soil moisture usually is reflected in the seasonal course of soil gas emissions (Schaufler et al., 2010). Authors describe difficulties when investigating the influence of a single climate param-25 eter from seasonal field measurements because confounding factors like N deposition, litterfall and nitrogen availability (Davidson et al., 2000;Pilegaard et al., 2006) co-vary or interact. With these confounding factors, measurements under natural field conditions cannot provide an unbiased estimate of the temperature sensitivity of SOM de-4437 composition (Kirschbaum, 1995). For field soil CO 2 fluxes, further complications arise from the contribution of autotrophic soil respiration (Schaufler et al., 2010). Laboratory incubations provide the best and least biased basis for estimating the temperature dependence of SOM decomposition (Kirschbaum, 2006). This assumption can be extended to the assessment of soil moisture dependence of heterotrophic soil respiration 5 (Schaufler et al., 2010). The combined effects of temperature and moisture changes are not necessarily additive (Beierkuhnlein et al., 2011;Larsen et al., 2011;Leuzinger et al., 2011). A two-factorial incubation design provides the opportunity to assess temperature and moisture effects independently and to investigate how the two climatic factors affect each other (Schaufler et al., 2010). 10 To investigate the combined effects of soil temperature and moisture on heterotrophic soil respiration from different land-use types, intact soil cores were taken from four representative land-use types from the ÉCLAIRE flux network, Europe and incubated in the laboratory under varying soil temperature and moisture levels. The main objectives of this study were (1) to determine the influence of soil temperature and moisture on 15 CO 2 efflux, (2) to calculate temperature and moisture sensitivities of CO 2 efflux coming from different land-use types, (3) to investigate the influence of moisture and landuse on temperature sensitivity of CO 2 efflux, and (4) to investigate the influence of temperature and land-use on moisture sensitivity of CO 2 efflux.

Study sites
Emission measurements of carbon dioxide under controlled conditions were conducted using soil monoliths from nine sites from the ÉCLAIRE flux network. Sites are located all over Europe; from the UK in the west to the Ukraine in the east; Italy in the south to Finland in the north. A list of all sites including relevant site information can be Introduction grasslands, arable lands and one peatland). Relevant soil characteristics are given in Table 2.

Sampling and experimental layout
Thirty-three undisturbed soil cores were collected at each of the investigation sites in spring 2012 after weekly-averaged soil temperatures reached 8 • C. This was done to 5 provide comparable conditions across sites with respective to sampling conditions. Soil cores were collected at 6 randomly distributed plots of approximately 10 m 2 within an overall area of approximately 50 m × 50 m at each site. Six soil samples were collected from each 10 m 2 plot at 6 spots. The upper 6 cm of the soil was collected in stainless steel cylinders (diameter, 7.2 cm; height, 7 cm). Soil cores were capped and sealed in 10 plastic bags to ensure original conditions and shipped in insulated coolers equipped with ice cartridges to our laboratory in Austria, where they were stored at 4 • C before being used for CO 2 flux measurements. 3 soil cores were used to determine gravimetric water contents. The gravimetric water content was determined for mineral soil by oven drying at 103 • C for three days to a constant weight. These water contents were 15 assumed to be representative for the rest of the soil samples from the same location, so that different water contents for the gas measurements could be established. The real gravimetric water content for each core was determined after gas flux measurements were completed. Intact soil cores were incubated in the laboratory for 22 h at the temperatures 5, 10, 20 15, 20, and 25 • C in a two factorial design of five soil moisture levels (5,20,40,60, 80 % water filled pore space, WFPS), before analysed for CO 2 fluxes. To design the experiment realistically moisture levels for the peatland site (UK-AMo) were set between 20-100 % WFPS. To reach the required moisture contents, distilled water was either added to too dry samples, or too moist samples were dried at 4 • C until they 25 reached the required moisture content. The lowest possible moisture content was 5 to 15 % WFPS for soil samples when drying at 4 • C. The second variable, soil temperature, was set by controlling the incubator to the desired temperature. Starting with 5 • C 4439 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | the temperature was increased every day in 5 • C steps up to an end temperature of 25 • C. We used 6 replicates for each moisture content in a complete factorial design in which each of the moisture contents was matched with each of the temperatures for all soil cores investigated. From three remaining cores soil characteristics (Table 2) were analyzed. Ammonium and nitrate was quantified according to Hood-Nowotny et al. (2010) using the ration 2.5 g soil : 25 ml KCL solution. Photometric analyses were conducted with a photometer from PerkinElmer® type 2300 EnSpire™. Conductivity was measured with a conducting meter 2F191 (WTW) and pH was measured in 0.01 m CaCl 2 , using the ratio 10 g soil : 25 mL CaCl 2 solution. The contents of total soil carbon (C t ) and nitrogen (N t ) 10 were determined with elemental analysis (NA-1500 Carlo Erba, Italy; ÖN1998).

Gas flux measurements
A fully automatic laboratory incubation system was used (Schindlbacher, 2004) to measure CO 2 flux rates. The system analysed CO 2 fluxes with an open flow system using a PP SYSTEMS WMA-2 (Amesbury, MA, USA) infrared CO 2 analyser. Twenty-four 15 modified Kilner jars were placed in a temperature-controlled incubator and connected to the instruments by Teflon tubes. Two of the chambers in the incubator were empty and served as control chambers for the gas measurements. The incubation chamber was flushed constantly with compressed ambient air (1.0 L min −1 ). The air sampling period in each test chamber was 6 min and of each reference chamber 4 min. A steady 20 state was achieved after approximately 4 min in the test chambers and 2.5 min in the reference chambers (Schindlbacher, 2004). Gas flux rates were calculated based on gas concentration changes over time according to Schindlbacher (2004) and mean values are shown with standard errors (SE).
To examine the temperature and moisture sensitivity of heterotrophic soil respi-Introduction   for temperature-CO 2 efflux relations and R(M) = R 0 + aM + bM 2 for moisture-CO 2 efflux relations.
To investigate how moisture content, temperature and land-use influence moisture sensitivity relative CO 2 values (relative to the CO 2 efflux of the lowest moisture content) were calculated to exclude the temperature contribution from the absolute CO 2 values. 5 Moisture sensitivity was calculated as the slope of a polynomial function of second degree which was fitted over the relative CO 2 values. This has been done for each temperature and site investigated.

Statistical analysis
Statistical analyses were performed with R (version 3.0.2) and SigmaPlot (version 11.0). Data were tested for normal distribution with the Shapiro-Wilk normality test and for variance homogeneity with the Constant Variance test. For multiple comparisons, the ANOVA test was performed to analyse significant differences. Significance of all tests was accepted at P levels < 0.05.

15
Intact soil cores from nine sites of the ÉCLAIRE flux network were incubated in the laboratory at the temperatures 5, 10, 15, 20, and 25 • C in a two factorial design of five soil moisture levels (5,20,40,60,80 (100) % water filled pore space, WFPS) before analysed for CO 2 fluxes. Data were normally distributed (Shapiro-Wilk test) and showed homogeneity of variances (Constant Variance test). CO 2 emissions differed 20 significantly among sites, temperatures and moisture contents (ANOVA). Comparison of CO 2 fluxes calculated as mean values over all temperature and moisture contents indicate that grassland sites showed the highest CO 2 emissions with 848.39 (±87.81) and 420.70 (±40.68) mg CO 2 −C m −2 h −1 for CH-Po and HU-BU, respectively, followed Introduction Spe where no significant moisture optimum could be detected. Additionally, a positive relation between CO 2 emissions and temperature is clearly visible.

Temperature sensitivity
The relationship between CO 2 emissions and temperature could be well described by a Gaussian model with the equation R(T ) = R 0 · e aT +bT 2  for all sites 10 investigated. Table 3 summarizes the fit of the Gaussian model for all sites investigated with all forest sites and the peatland site ranging between an R 2 of 0.990 and 1; grasslands between an R 2 of 0.871 and 1; and arable lands between an R 2 of 0.639 and 1.
The temperature course of individual soil cores exhibited a good fit to the Gaussian model. As a result mean values of CO 2 fluxes for each of the five temperatures per 15 moisture content and site were calculated to fit the equation. When taking all samples R 2 ranged between 0.019 (UA-Pet; 6 % WFPS) and 0.958 (NL-Spe; 30 % WFPS) due to the variability between soil cores. Based on the Gaussian model temperature sensitivities were calculated as Q 10 values from 5-15 • C for each moisture content and site investigated. Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Q 10 at 15 • C showed that temperature sensitivity converged towards 2 as temperature was increasing. There was no positive or negative relationship of temperature sensitivities to increasing moisture content. However, the variability between Q 10 values of different moisture contents (highest Q 10 value minus lowest Q 10 value illustrated in Table 3 for 5 and 15 • C) at a certain temperature decreases with increasing temperature. at NL-Spe, Q 10 (5 • C) = 7.78 at FI-Hyy) which can also be seen in Table 3.

Moisture sensitivity
Moisture sensitivity (MS) was calculated as the slope of a quadratic function fitted over relative CO 2 values (to exclude the temperature contribution). Figure 3a shows relative values calculated for a deciduous forest in Italy (IT-IFo) and Fig. 3b shows the quadratic function fitted over relative CO 2 emissions for the same forest at 5 • C. tionally moisture sensitivities of CO 2 fluxes coming from arable lands showed a positive relationship to temperature which can be also seen in Table 4, namely that moisture sensitivities at 5 % WFPS increased with temperature for both arable lands, FR-Gri and UA-Pet.

5
Land-use generally had a substantial influence on carbon dioxide fluxes, with the order of CO 2 emission rates of the different land-use being grassland > peatland > forest/arable land (P < 0.001) which is in line with observations by Schaufler et al. (2010); Raich and Tufekciogul (2000); Ambus and Robertson (2006). Heterotrophic soil respiration responded strongly to varying temperature and moisture content (Ferréa et al.,  , 1995). Additionally results showed a positive correlation between CO 2 emissions and temperature (Davidson et al., 1998;Luo et al., 2012;Wang et al., 2006;Wu et al., 2010). In agreement with other studies the relationship between CO 2 emissions and temperature could be well described by a Gaussian model with the equation Tuomi et al., 2008;Vanhala et al., 2008 (Kirschbaum, 1995;Lloyd and Taylor, 1994;Luo et al., 2001) that temperature sensitivity is negatively correlated to temperature which was true for most of the moisture contents and sites investigated (except one arable land UA-Pet and IT-BFo at 26 % WFPS, NL-Spe at 18 % WFPS; CH-Po at 5 % WFPS and UK-AMo at 83 % WFPS). Additionally temperature sensitivity converged towards 2 as temperature increased for all moisture contents at all sites investigated. We found that precipitation can influence temperature sensitivity of CO 2 efflux due to the decrease of the variability between Q 10 values of different moisture contents (highest Q 10 value minus lowest Q 10 15 value) at each moisture point with increasing temperature. At low temperatures Q 10 values vary more between dry and wet conditions. At higher temperatures the effect of water and temperature on Q 10 is very low as Q 10 converges towards 2. Additionally to the Gaussian model equation we applied the Arrhenius function (R(T ) = R 0 · e aT −1 ) to our results which showed similar trends but unrealistic Q 10 values at temperatures 20 below 8 • C (Q 10 ranging between 20 and 2000).
Our results showed that no distinct relationship (neither positive nor negative) could be found between temperature sensitivity and moisture content at any of the investigated sites. Janssens and Pilegaard (2003) and Qi et al. (2002) expected a positive relationship between temperature sensitivity and moisture content due to the assump- positive relationship between temperature sensitivity and soil moisture content in their field measurements. Peng et al. (2009) described in a review of 52 papers a negative correlation between Q 10 values and mean annual precipitation. However, all these conclusions were achieved through seasonal field measurements at which derivation of the influence of a single climate parameter is difficult, because of incorporated seasonal changes in root biomass, litter inputs, microbial population, nitrogen availability and other seasonally fluctuating processes and conditions, and thus reflect community responses, which may differ from temperature and moisture responses of the respiratory processes (Davidson et al., 2000;Janssens and Pilegaard, 2003;Pilegaard et al., 2006) and can even be partly decoupled from actual soil temperature and mois-  2004) presented a case study of how the seasonal Q 10 of soil respiration can be decoupled from the temperature sensitivity of soil respiration indicating that the large differences in seasonal Q 10 do not represent differences in the temperature sensitivity of the 15 soil metabolism. Kirschbaum (2006) and Lützow and Kögel-Knabner (2009) considered that laboratory incubations provide the best and least biased basis for estimating the temperature sensitivity of organic matter decomposition. This assumption can be extended to the assessment of soil moisture sensitivity of organic matter decomposition (Schaufler et al., 2010). Another laboratory incubation study by Schindlbacher 20 et al. (2007) showed that different soil moisture contents of trenched and control plots affected rates of heterotrophic soil respiration, but did not affect the temperature sensitivity of heterotrophic respiration which is in agreement with our results. We found at both the coniferous forest sites a strong increase of the temperature sensitivity at a moisture range between 20-40 % WFPS. At coniferous sites the amount of 25 recalcitrant material is higher (Landsberg and Gower, 1997;Wang et al., 2006) than at all other sites investigated. Temperature sensitivity of soil respiration increases with substrate recalcitrance as long as environmental constraints are not limiting decomposition (Conant et al., 2008;Hartley and Ineson, 2008;Karhu et al., 2010 -Knabner, 2009;Zimmermann and Bird, 2012) because of the higher number of steps needed for decomposition of more complex substrates. Also according to kinetic theory the temperature sensitivity of decomposition increases with increasing molecular complexity of the substrate due to higher activation energy of recalcitrant substrate (Hartley and Ineson, 2008;Vanhala et al., 2008). We hypothesize that a moisture range 5 between 20-40 % WFPS promotes decomposition of recalcitrant material in coniferous forests. Not the absolute amount of carbon dioxide increases at this moisture range as NL-Spe shows no significant CO 2 maximum at any moisture content and Hyytiälä has its maximum between 40-70 % WFPS. Results rather state that within this moisture range recalcitrant material is being favourably decomposed to easy degradable mate-10 rial. Initially discriminative differences in Q 10 values between moisture contents evened out with increasing temperature as Q 10 values converged towards 2 for all moisture contents. We couldn't see any obvious trends of Q 10 values among land uses which is in agreement with Wu et al. (2010). Peng et al. (2009) found differences of Q 10 values 15 among ecosystem types but did compare field measurements and different temperatures which both result in different Q 10 values (Curiel Yuste et al., 2004;Janssens and Pilegaard, 2003;Kirschbaum, 1995;Lloyd and Taylor, 1994;Luo et al., 2001;Schindlbacher et al., 2009). 20 In our study moisture sensitivity was calculated as the slope of a polynomial function of second degree. The use of quadratic functions for the description of the relationship between heterotrophic soil respiration and moisture content is widely common (Moyano et al., 2013;Rodrigo et al., 1997). Our results show that significant moisture effects (P < 0.05) occurred only at higher temperatures which is in agreement with other stud-25 ies (Teepe et al., 2004;Wu et al., 2010). To calculate the moisture sensitivity without temperature influence we took relative CO 2 values for regression analysis to exclude the temperature contribution. 4447

Moisture sensitivity
Many articles can be found on the topic of temperature sensitivity. However, much fewer articles calculate moisture sensitivities. Our results indicate that moisture sensitivity is highest at very dry and wet conditions. Results by Moyano et al. (2012) indicate that moisture sensitivity is negatively correlated to soil moisture. However, Moyano et al. (2012) calculated moisture sensitivity as the proportional response of soil 5 microbial respiration to a 0.01 increase in soil moisture of a certain unit. As CO 2 values decline after a moisture optimum this mathematical approach results in moisture sensitivities showing highest values at dry conditions decreasing progressively with increasing moisture content until converging to a certain value. Our approach to calculate moisture sensitivities indicates that moisture sensitivities decrease until reaching 10 the moisture optimum and increase again after that (negative values after reaching the optimum moisture content (MC opt ) only indicate a decrease of CO 2 emissions with increasing moisture content; positive values indicate an increase of CO 2 emissions with increasing moisture content). Therefor we can show that changing moisture content has a higher impact on CO 2 emissions at dry and wet conditions than at intermediate 15 moisture conditions.