Temperature sensitivity of microbial respiration, nitrogen mineralization, and potential soil enzyme activities in organic alpine soils

Authors


Abstract

[1] Investigations focusing on the temperature sensitivity of microbial activity and nutrient turnover in soils improve our understanding of potential effects of global warming. This study investigates the temperature sensitivity of C mineralization, N mineralization, and potential enzyme activities involved in the C and N cycle (tyrosine amino-peptidase, leucine amino-peptidase, ß-glucosidase, ß-xylosidase, N-acetyl-ß-glucosaminidase). Four different study sites in the Austrian alpine zone were selected, and soils were sampled in three seasons (summer, autumn, and winter). A simple first-order exponential equation was used to calculate constant Q10 values for the C and N mineralization over the investigated temperature range (0–30°C). The Q10 values of the C mineralization (average 2.0) for all study sites were significantly higher than for the N mineralization (average 1.7). The Q10 values of both activities were significantly negatively related to a soil organic matter quality index calculated by the ratios of respiration to the organic soil carbon and mineralized N to the total soil nitrogen. The chemical soil properties or microbial biomass did not affect the Q10 values of C and N mineralization. Moreover, the Q10 values showed no distinct pattern according to sampling date, indicating that the substrate quality and other factors are more important. Using a flexible model function, the analysis of relative temperature sensitivity (RTS) showed that the temperature sensitivity of activities increased with decreasing temperature. The C and N mineralization and potential amino-peptidase activities (tyrosine and leucine) showed an almost constant temperature dependence over 0–30°C. In contrast, ß-glucosidase, ß-xylosidase, and N-acetyl-ß-glucosaminidase showed a distinctive increase in temperature sensitivity with decreasing temperature. Low temperature at the winter sampling date caused a greater increase in the RTS of all microbial activities than for the autumn and summer sampling dates. Our results indicate (1) a disproportion of the RTS for potential enzyme activities of the C and N cycle and (2) a disproportion of the RTS for easily degradable C compounds (ß-glucose, ß-xylose) compared with the C mineralization of soil organic matter. Thus temperature may play an important role in regulating the decay of different soil organic matter fractions due to differences in the relative temperature sensitivities of enzyme activities.

1. Introduction

[2] Alpine soils store a large pool of soil organic matter [Körner, 1999]. The alpine zone is adapted to a cool environment characterized by a long seasonal snow cover and a short vegetation period. A recent study predicted an increase of mean summer temperature between 3 and 5°C for the next 100 a, being more intensive for the European Alps than for temperate regions in Europe [Heimann and Sept, 2000]. The strong influence of temperature on carbon mineralization has been reviewed for different conditions [e.g., Kirschbaum, 1995; Kirschbaum, 2000; Lloyd and Taylor, 1994]. The hypothesis that respiration depends more on temperature than net photosynthesis is crucial for the ecosystem C balance [Kirschbaum, 2000]. This makes investigations on the temperature dependency of C mineralization very important. The temperature quotient (Q10) describes the ratio of activity for a 10°C (or 10 K) change of temperature. However, there is still no consensus about the form of the relationship between temperature and decomposition [Kätterer et al., 1998]. The temperature quotient (Q10) is often calculated by simple first-order exponential functions yielding a constant Q10 over the investigated temperature range. Q10 values of soil respiration are influenced by substrate quality [Fierer et al., 2006; Mikan et al., 2002]. On the other hand, flexible exponential functions implying greater temperature sensitivity at low temperatures have been found to best describe the temperature response for C or N mineralization over a wide temperature range [Kirschbaum, 1995; Lloyd and Taylor, 1994].

[3] The temperature dependency of N mineralization is more difficult to investigate because of methodological problems in distinguishing between different soil N turnover processes. In general, the temperature sensitivity of net N mineralization appeared to be slightly lower than of C mineralization for a wide range of investigated soils [Kirschbaum, 1995]. Kirschbaum [1995], however, pointed out that the high individual range of reported values precludes a final conclusion. Magid et al. [2001] found that the decay of plant residues at low temperatures resulted in a higher proportion of mineralized nitrogen compared with C mineralization. The net N mineralization is a result of the mineralization-immobilization turnover (MIT). There was no relationship between net N mineralization rate of various humus samples and incubation temperatures below 15°C [Niklińska et al., 1999]. The 15N pool dilution technique showed a higher temperature dependency of gross N immobilization versus gross N mineralization for soils being treated with green manure [Andersen and Jensen, 2001]. This temperature-dependent decoupling of MIT may be crucial in understanding the temperature dependence of N cycling of easily degradable substrates like plant residues in agroecosystems. However, little information is available about the temperature sensitivity of C and N processes in organic soils in alpine regions. Additionally, the temperature sensitivity of soil microbial activities in the alpine environment may change seasonally as an adaptation to seasonal changes in microbial biomass [Lipson et al., 2002].

[4] Extracellular enzymes are important because they catalyze the rate-limited steps of decomposition and nutrient cycling [Sinsabaugh, 1994]. Surprisingly, there are very few published data about temperature sensitivities of a wider range of potential enzyme activities within a given soil [Trasar-Cepeda et al., 2007]. In general, potential enzyme activities have been found to depend less on temperature compared to the C mineralization with Q10 values < 2 [Browman and Tabatabai, 1978; Tabatabai, 1982].

[5] This study investigates the C and N mineralization as well as the potential activity of soil enzymes involved in C and N cycling, which was investigated over a temperature range of 0–30°C. We tested the hypothesis that (1) the Q10 value of C mineralization is higher than that of N mineralization, (2) Q10 values of C and N mineralization are lower in winter than in summer as an adaptation of the microbial biomass, (3) the relative temperature sensitivity (RTS) of C and N mineralization as well as enzyme activity increases with decreasing temperature, and (4) the quality of soil organic matter influences the temperature sensitivity of C and N mineralization. Four alpine study sites were selected (one meadow and three fen plots), and soil samples were taken in three seasons (summer, autumn, and winter).

2. Material and Methods

2.1. Study Sites Description

[6] The study sites lie in the Rotmoos valley of the Ötztal range (46°50′N, 11°03′E) in Tyrol (Austria) above the present tree line (2250 m above sea level). The snow-free period is about 4.5 months (June to mid-October) (M. Strobel, personal communication, 2003), mean annual precipitation is 820 mm (1970–1996) and mean annual air temperature (1997–1998) is −1.3°C [Kaufmann, 2001].

[7] Four study sites were chosen along a natural moisture transect (one alpine dry meadow and three plots in the fen Rotmoos); they differed in water balance and plant community. The first site is a well-drained alpine dry meadow (meadow) classified as a Curvulo-Nardetum (Zeltner, personal communication, 2003). The other three sites are located in the Rotmoos fen. The fen is mainly fed by surface water from runnels flowing down the hill slopes of the flanking mountains. The first site in the Rotmoos fen (transitional) is a Carici echinatae-Trichophoretum caespitosi community [Rybníček and Rybníčková, 1977]. This transitional site is located closer to the meadow site than the other fen sites and contains a high cover of plant species belonging also to the Curvulo-Nardetum. The second fen site (fen) is located deeper in the Rotmoos fen and is characterized by a typical Carici echinatae-Trichophoretum caespitosi plant community [Rybníček and Rybníčková, 1977]. The species richness was lower than at the transitional site. The third fen site (wet fen) is located in the centre of the Rotmoos fen and consists solely of Carex nigra L. with a sparse vegetation cover compared to all other study sites. This site can be temporarily flooded by ground water.

2.2. Soil Sampling, Chemical Soil Properties, and Microbial Biomass (Cmic)

[8] All soils represent mature alpine stages and have been developed since the last ice age (5000–10,000 a ago) [Bortenschlager, 1970]. The parent material of the meadow soil consists of loamy sandy silicates and granite with a maximum soil depth of 50 cm. The substrate is silty sand, and the soil is classified as an acid Cambisol. All soils in the fen are rheic Histosols with soil depths deeper than 1 m. The sites were randomly sampled in 0–5 cm soil depth (n = 5 for each study site) at three sampling dates: The first sampling date was in autumn (a) (3 October 2002), the second date was in summer (s) (5 July 2003), and the third date was in winter (w) (2 February 2004) under a snow cover of about 180 cm. The mean daily soil temperature in 5 cm depth for ±2 weeks around the sampling date was similar for all study sites. The overall order was s (14.2°C) > a (3.9°C) > w (0.0°C). The samples were carefully homogenised by hand and the roots were sorted out. The homogenised soil was stored frozen (−20°C) in plastic bags until analysis. The chemical soil properties image Corg, Nt) and microbial carbon (Cmic) were determined for every sampling date. The image was potentiometrically determined using 5 g soil and 10 mL 0.01 M CaCl2 solution. The total carbon (Ct) and total nitrogen (Nt) were determined on air-dried soil by dry combustion at 1200°C using a LECO CNS 2000 element analyser with infrared and thermal conductivity detector. Since all soils were acidic, Ct was considered as soil organic carbon (Corg). Cmic was determined from field-moist soil samples with the chloroform fumigation-extraction (CFE) method. The soil samples (1 g, n = 3) were fumigated for 1 d in chloroform vapor, followed by extraction (0.5 M K2SO4), filtration and analysis of the dissolved organic carbon (DOC) by flow injection analysis (Dimatoc 100 TOC/TN-analyzer, Dimatec, Germany). The DOC of blanks (n = 3) were extracted from each soil sample without fumigation. Cmic was calculated by the difference between fumigated samples and the blanks with the recovery factor of 0.45 proposed for peat soils [Sparling et al., 1990]. The chemical properties and Cmic of the autumn samples are listed in Table 1. There were no significant differences between the sampling dates for any soil property at any study site (F(date) < 2.9; p > 0.06, two-way ANOVA).

Table 1. Chemical Soil Properties and Cmic of the Study Sites (Autumn Sampling)a
 pH(CaCl2)Corg, %Nt, %Corg Nt−1, % %−1Cmic Corg−1, mg g C−1
  • a

    Different letters in the column indicate significant differences of the means between study sites. Values in parentheses are standard errors.

Meadow3.9 (0.1)a17.0 (1.9)a1.0 (0.1)a17.1 (1.6)a42.8 (6.2)a
Transitional4.3 (0.1)a28.2 (1.8)b1.6 (0.1)b17.2 (0.7)a23.3 (4.1)b
Fen4.3 (0.1)a30.8 (2.2)b1.5 (0.1)b21.1 (0.8)b17.2 (3.8)b
Wet fen4.0 (0.1)a35.1 (1.3)b1.6 (0.1)b22.0 (0.9)b18.9 (5.2)b

2.3. Temperature Incubation Experiment

[9] C mineralization, N mineralization and enzyme activities involved in C and N cycling were measured in all soil samples over a temperature range of 0–30°C (5°C steps). Two incubators (Heraeus BK 600) and a climate chamber (SBS C120) were used to regulate the temperature with a precision (±1°C) that prevented any sample in any analysis from freezing during the incubation time at 0°C.

2.3.1. C Mineralization

[10] The frozen soil samples were allowed to thaw at 4°C for 10 d before the incubation experiment started. The water content of the soil samples was not adjusted prior the incubation experiment to avoid disturbance and/or adaptation to surrounding temperature during time of adjustment. However, the water content of the soils was not expected to limit soil respiration; all samples corresponded to a range of 79 to 98% maximum water holding capacity. Temperature sensitivity was determined by measuring the activity response of a soil sample to the different temperature steps. Soil samples (10 g) were incubated in airtight flasks and the CO2 evolved was trapped in 2 mL of NaOH (1 M) and titrated by HCL (0.1 M) after treated with 0.5 mL saturated BaCl2. Blanks (n = 3) were run in the same procedure for each incubation temperature. The soil samples were allowed to adapt to the incubation temperature in open flasks for at least 2 h before the measurement started. The incubation time varied depending on incubation temperature, starting with 9 d at 0°C and ending with 1 d at 30°C (in total, 25 d). During the experiment, soil moisture was kept constant. A preliminary experiment showed that soil respiration was constant for all study sites after 7 d and that cumulative CO2 evolution (at 20°C) for all study sites was linear (R2 > 0.98) for at least 6 weeks (data not shown). We therefore conclude that the temperature sensitivity during the total incubation time is not affected by decreasing respiration rates or by changes in the substrate quality of the soil organic matter.

2.3.2. N Mineralization

[11] The N mineralization was measured under waterlogged anaerobic conditions according to Keeney [1982] and Kandeler [1996]. Thus three subsamples (5 g) of each soil sample were mixed with 15 mL deionized water and incubated for 4 d. The N-NH4+ produced was measured colorimetrically (Berthelot color reaction) at 660 nm after being extracted with KCL (2 M) and filtered (N-free filters). Blanks were run in triplicates for each soil sample (immediately extracted and measured without incubation). The N mineralization activity was determined as the difference between samples and blanks (zero-order reaction). A preliminary study was conducted to investigate the effect of incubation time (2, 4, 6, 8, and 12 d) on NH4+ production at different incubation temperatures (0°C, 15°C and 30°C) for the summer soil samples. For all soils and all incubation temperatures, the NH4+ produced was constant for at least 4 d. After this period, values fell or were even slightly negative (probably indicating N immobilization) (data not shown). Waterlogging of soil samples may not immediately result in anaerobic soil conditions, especially at low incubation temperatures with their usually low activity. N mineralization may therefore be underestimated because ammonium is oxidized during incubation. The potential nitrification activity of the study soils was tested according to Berg and Rosswall [1985] and no significant nitrification activity was detected. Thus ammonium oxidation is not important for these acidic alpine soils and does not influence the N mineralization rates in our study.

2.3.3. Enzyme Activity

[12] Hydrolytic enzymes involved in C and N processes were investigated using fluorimetric microplate assay [Marx et al., 2001]. The activities of ß-glucosidase (GLUC; E.C. 3.2.1.21), ß-xylosidase (XYL; E.C. 3.2.2.37) and N-acetyl-ß-glucosaminidase (NAG; E.C. 3.2.1.30) were measured using 4-methylumbelliferyl (MUB)-ß-1,4-glucoside, MUB-ß-xylopyranoside and MUB-N-acetyl-ß-1,4-glucosaminide, respectively. The activity of leucine amino-peptidase (LEU; E.C. 3.4.11.1) and tyrosine amino-peptidase (TYRO; E.C. 3.4.11) were measured using L-leucine-7-amino-4-methyl-coumarin and L-tyrosine-7-amido-4-methyl-coumarin, respectively. The enzyme activity was determined as the hydrolytic cleavage of 4-methylumbelliferone (MUB) and 7-amido-methylcoumarin (AMC) from MUB- and AMC-labelled substrate, respectively. Before measuring the enzyme activity, suspensions (soil:water = 1:100) were prepared from each soil sample using low-energy sonication (50 J s−1 output energy for 2 min according to Stemmer et al. [1998]) and 20 μl was transferred into vials of a 96-well microplate. 80 μl of buffer (100 mM 2-(N-morpholino)ethanesulfonic acid buffer (pH 6.1) for MUB substrate or TRIZMA buffer (pH 7.8) for AMC substrate and 100 μl of substrate (1 mM) were added to each sample and incubated at the respective temperature using incubators or the climate chamber. All solutions and soil suspensions were prepared with autoclaved deionized water. Impurities of the solutions and/or the stability of the substrates were proved with blanks (samples without soil suspension). All temperature steps were conducted with each soil solution within 12 h.

[13] The release of MUB or AMC was measured using an automated luminescense spectrophotometer (FLx 800 microplate, Bio-Tex Instruments, Inc.; emission 446 nm, extinction 377 nm) at 20 to 40 min intervals for at least 2 h (depending on incubation temperature). After each measurement (∼1 min) the microplates were immediately returned to the incubators or climate chamber. The produced fluorescense was converted into amounts of MUB or AMC with soil-specific standards to correct for possible quenching effects of the soil matrix on the fluorescense intensity.

2.4. Temperature Models

[14] Two different temperature models were used to calculate the temperature sensitivity of the investigated activities. The first model is a simple first-order exponential equation (1):

equation image

A is the exponential constant, or activity at 0°C, and B the exponential parameter of the equation.The second model is a flexible exponential equation (2) [Lloyd and Taylor, 1994]:

equation image

A1 is the exponential constant, while B1 (loosely related to the activation energy) and T0 are the exponential parameters [see Kirschbaum, 2000; Lloyd and Taylor, 1994]. This function implies a temperature dependence decreasing with increasing temperature and was found to be safely applicable for soil respiration at a temperature range of −10°C to almost 40°C [Lloyd and Taylor, 1994]. The relative temperature sensitivity (RTS) was calculated according to equation (3):

equation image

Equation (3) is a derivation function of equation (2) and allows the temperature sensitivity to be calculated for a given temperature [Lloyd and Taylor, 1994].

[15] Equation (1) was used to calculate constant Q10 values of the C and N mineralization. Q10 values of the potential enzyme activities were not calculated because the hydrolytic cleavage of MUB- or AMC substrate may bias the activation energy and therefore the temperature dependency [Fey and Conrad, 2003]. Hence the mutated activation energy of the artificial substrates influences the absolute temperature sensitivity (according to the Arrhenius theory) and is therefore not useful in interpreting these values.

[16] Equation (2) was used to calculate the RTS of the activities (C mineralization, N mineralization, potential enzyme activities) for two different temperature frames (Δ 5°C). We select the temperature frames 0–5°C: RTS(0–5) and 13–18°C: RTS(13–18), which incorporated the average mean daily soil temperatures (±2 weeks) at the three sampling dates for all study sites. Since RTS at a given temperature depends on the activation energy, the ratio RTS(0–5)/RTS(13–18) was calculated. This procedure allows comparing the RTS of all investigated activities. Hence a higher RTS(0–5)/RTS(13–18) ratio indicates a greater increase of the RTS with decreasing temperature than a lower RTS(0–5)/RTS(13–18).

2.5. Data Handling and Statistics

[17] All results are calculated on an oven-dried weight basis (105°C, 2 d). The C mineralization (μg C-CO2 g−1 C d−1) was calculated on a Corg basis. This relation is an index for substrate quality [e.g., Fierer et al., 2006; Mikan et al., 2002]. Higher respiration rates related to the Corg indicate a better substrate quality. The N mineralization (μg N-NH4+ g−1 N d−1) was calculated on a Nt basis. The potential enzyme activities (nmol MUB g−1 min−1 or nmol AMC g−1 min−1) were related to the bulk soil. The Q10 values were linearly correlated with the chemical soil properties (pH, Corg, Nt, Corg Nt−1), Cmic and the exponential constant A derived by equation (1) (AC: exponential constant A of C mineralization; AN: exponential constant A of N mineralization) in order to investigate the influence of these parameters on the temperature sensitivity. An a priori inverse autocorrelation between A and Q10 values (as demonstrated by Reichstein et al. [2005] when only one individual data set was used) did not exist because our A and Q10 values are derived from more individual data sets (n = 60 for each activity). Normal distribution of the data was tested by the Kolmogorow-Smirnoff-goodness-of-fit test, and homogeneity of variances was tested by Levene's test. Differences of the Q10 values (C and N mineralization) and the RTS(0–5)/RTS(13–18) ratios (C, N mineralization and enzymes (GLUC, XYL, NAG, TYRO, LEU)) were tested by a simple two-factorial analysis of variance (factor 1: sampling date (Fdate); factor 2: study site (Fsite)). Differences of the mean Q10 values between the C and N mineralization were tested with a mixed linear model using the restricted maximum likelihood (REML) measure. The model contains a fixed three-factorial design (factor 1: element of mineralization (C or N) Felement, 2: Fdate, and 3: Fsite) combined with a repeated random subject term (date × site × replicate (within the site × date interaction)). This random interaction term accounts for the correlation of both activities within soil samples. Differences of the RTS(0–5)/RTS(13–18) between activities (C and N mineralization and enzymes (GLUC, XYL, NAG, TYRO, and LEU)) were tested with the same mixed linear model design (factor 1: activity (Factivity), 2: Fdate, 3: Fsite, and the repeated interaction of date × site × replicate (within the site × date interaction)). All values in the figures and tables are means (± standard error). For ease of interpretation, all temperatures in the figures are given in the Celsius scale.

3. Results

3.1. Q10 of the C and N Mineralization

[18] The simple first-order exponential equation was used to calculate constant Q10 values for the entire investigated temperature range (0–30°C) (Table 2). The predicting ability (R2) of the fitted model for the individual field replicates was 0.85 to 0.99 (C mineralization) and 0.78 to 0.96 (N mineralization). The mean (±SE) Q10 value of the C mineralization for all field replicates, study sites and sampling dates was 2.0 (±0.04). There was only a small significant difference for sampling date (Fdate = 4.3, p = 0.019; two-way ANOVA), with a higher mean Q10 value in autumn (2.1) than in summer (1.9) and winter (2.0). The mean (±SE) Q10 value of the N mineralization for all field replicates, sampling date and study sites was 1.7 (±0.03). Significant differences were found for sampling date (Fdate = 7.1, p = 0.002; two-way ANOVA), with a higher mean Q10 value in winter (1.8) than in summer (1.6) and autumn (1.6). The study site (Fsite) or the factor interaction Fsite × Fdate were not significant for C mineralization (p ≥ 0.114) and N mineralization (p ≥ 0.208). The mixed model identified an overall (all sampling dates and study sites) higher mean Q10 value for C mineralization than for N mineralization (Felement = 37.8, p < 0.001). However, there was also a small significant Felement × Fdate interaction (Felement × date = 5.2, p = 0.007), mirroring the different results of the C and N mineralization for Fdate in the two-way ANOVA. In addition, a small significant difference was found for sampling date (Fdate = 5.8, p = 0.004), with a lower mean Q10 value in summer (1.7) than in autumn (1.9) and winter (1.9). The factor study site (F(site)) and all other factor interactions were not significant (p ≥ 0.112).

Table 2. Model Parameters of the Simple First-Order Exponential Function (Equation (1)) for C and N Mineralizationa
 Simple First-Order Model, f(T) = A · e(BT); T, °C
C MineralizationN Mineralization
ACBQ10R2ANBQ10R2
  • a

    Units are AC: (μg C-CO2 g−1 C d−1), AN: (μg N-NH4+ g−1 N d−1), and B: (°C−1). Values are means (± standard errors).

Summer        
   Meadow49.3 (9.1)0.059 (0.004)1.8 (0.1)0.87–0.9473.6 (10.1)0.048 (0.003)1.6 (0.1)0.83–0.95
   Transitional36.3 (6.2)0.054 (0.006)1.7 (0.1)0.91–0.9882.3 (9.3)0.039 (0.004)1.5 (0.1)0.86–0.93
   Fen34.1 (7.3)0.063 (0.004)1.9 (0.1)0.85–0.9394.7 (8.2)0.044 (0.004)1.5 (0.1)0.82–0.92
   Wet fen22.9 (6.7)0.068 (0.005)2.0 (0.1)0.87–0.9851.5 (11.6)0.053 (0.005)1.7 (0.1)0.83–0.91
Autumn        
   Meadow53.0 (9.9)0.073 (0.004)2.1 (0.1)0.88–0.99103.6 (14.3)0.041 (0.005)1.5 (0.1)0.84–0.96
   Transitional27.4 (5.1)0.076 (0.005)2.1 (0.1)0.91–0.9755.0 (6.1)0.054 (0.004)1.7 (0.1)0.84–0.94
   Fen34.1 (7.1)0.070 (0.005)2.0 (0.1)0.88–0.9675.7 (5.3)0.041 (0.005)1.5 (0.1)0.81–0.91
   Wet fen24.4 (4.1)0.081 (0.004)2.3 (0.1)0.89–0.9787.0 (8.4)0.043 (0.003)1.5 (0.1)0.78–0.90
Winter        
   Meadow95.3 (16.1)0.056 (0.005)1.7 (0.1)0.89–0.95144.6 (17.3)0.047 (0.003)1.6 (0.1)0.84–0.95
   Transitional17.4 (3.9)0.078 (0.004)2.2 (0.1)0.87–0.9758.4 (7.5)0.061 (0.004)1.8 (0.1)0.85–0.94
   Fen48.5 (5.1)0.057 (0.005)1.8 (0.1)0.89–0.9664.0 (8.3)0.061 (0.004)1.8 (0.1)0.80–0.90
   Wet fen20.2 (4.2)0.076 (0.004)2.1 (0.1)0.88–0.9738.8 (5.4)0.056 (0.004)1.7 (0.1)0.78–0.92

[19] The correlation of the Q10 values with the exponential constant A (index for substrate quality) was examined to investigate the influence of substrate quality on the temperature sensitivity of C and N mineralization. The index for substrate quality of the C mineralization (AC) and N mineralization (AN) was inversely related to the corresponding Q10 values (Figure 1). The regressions explained 44% (C mineralization) and 54% (N mineralization) of the variability of Q10 values. The slopes of the linear regressions and the y axis intercepts were almost similar, being slightly lower for the C than for N mineralization. There were no significant correlations between Q10 values and all other independent soil properties (pH, Corg, Nt, Corg Nt−1 and Cmic (p > 0.11)).

Figure 1.

Relationship between Q10 values and the exponential constant A (index of soil organic matter quality). Q10 values, A of C mineralization (AC), and N mineralization (AN) were obtained from a simple first-order exponential equation (equation (1)).

3.2. Substrate Quality Index of the C and N Mineralization

[20] The substrate quality index of the C mineralization (AC) was related to the substrate quality index of the N mineralization (AN) (Figure 2). If both soil-derived activities were in quasi equilibrium with the available soil substrate (Corg, Nt), we would expect a 1:1 dialog. A significant linear relationship between AC and AN was found, being almost parallel to the 1:1 line. The regression line, however, did not go through the origin. If the regression line was forced to go through the origin, the relationship between AC and AN was not significant. Thus both activities were much better described by including a constant y axis intercept term (41.6) in the regression equation. This leads to the assumption that the different methods employed here to study C and N mineralization may use different pools of soil organic matter and are not directly related.

Figure 2.

1:1 dialog between the exponential constant A (index of soil organic matter quality) of C mineralization (AC) and N mineralization (AN); obtained from a simple first-order exponential equation (equation (1)). Solid line is the regression line between C and N mineralization; dashed line represents the regression through origin.

3.3. Relative Temperature Sensitivity (RTS) of the C and N Mineralization and Enzyme Activities (TYRO, LEU, GLUC, XYL, NAG)

[21] The flexible temperature dependence model [Lloyd and Taylor, 1994] was used to determine the relative temperature sensitivity (RTS) of all activities (Tables 3 and 4). The equation follows the hypothesis that temperature sensitivity increases with decreasing temperature. The predicting ability (R2) of the fitted flexible model for the individual field replicates of all activities (all sampling dates and study sites) was between 0.78 and 0.96. Its predicting ability was similar to the simple first-order exponential equation model for C and N mineralization (Tables 2 and 3). RTS values (C and N mineralization, enzyme activities) were plotted against temperature in Figure 3. The C and N mineralization and the amino-peptidases (TYRO, LEU) showed almost no increase of the RTS with decreasing temperature. In contrast, GLUC, XYL and NAG more closely followed the logic of the flexible model, with much higher RTS values at low than at higher temperatures. This relative trend of the curves was generally observed for all study sites and all sampling dates (data not shown).

Figure 3.

Relative temperature sensitivity (RTS) of the investigated soil microbial activities as a function of temperature (equation (3)) (study site: alpine meadow; sampling date: summer); C mineralization (Cmin), N mineralization (Nmin), and enzyme activities: Tyrosine amino-peptidase (TYRO), leucine amino-peptidase (LEU), ß-glucosidase (GLUC), ß-xylosidase (XYL), and N-acetyl-ß-glucosaminidase (NAG).

Table 3. Model Parameters of the Flexible Temperature Dependence Function (Equation (2)) for C and N Mineralizationa
 Flexible Model [Lloyd and Taylor, 1994], f(T) = A1 · image
C MineralizationN Mineralization
A1B1T0R2A1B1T0R2
  • a

    Units are A1: (μg C-CO2 g−1 C d−1) (C mineralization) and (μg N-NH4+ g−1 N d−1) (N mineralization) and B1, T0: (K). Values are means (± standard errors).

Summer        
   Meadow3.12 108 (4.23 107)3853 (582)56 (9)0.88–0.948.92 106 (1.12 106)2553 (306)88 (13)0.84–0.95
   Transitional5.68 108 (5.24 107)4216 (564)46 (12)0.91–0.986.78 105 (7.58 104)1720 (196)114 (15)0.85–0.93
   Fen2.24 107 (2.80 106)3284 (539)63 (6)0.85–0.933.24 105 (4.35 104)1357 (171)137 (19)0.82–0.92
   Wet fen8.46 106 (6.18 105)2841 (432)87 (14)0.87–0.976.17 104 (5.68 103)1167 (132)143 (17)0.85–0.91
Autumn        
   Meadow4.27 107 (5.14 106)2887 (349)87 (16)0.87–0.982.89 105 (3.96 104)1207 (140)149 (24)0.84–0.97
   Transitional2.84 106 (3.18 105)2353 (326)99 (13)0.91–0.975.88 106 (8.35 105)2667 (253)79 (11)0.84–0.94
   Fen4.46 107 (6.42 106)3221 (432)73 (10)0.87–0.966.48 105 (8.22 104)1494 (186)137 (26)0.81–0.92
   Wet fen3.13 106 (2.53 105)2479 (335)96 (18)0.89–0.981.38 105 (2.26 104)1124 (152)151 (20)0.78–0.90
Winter        
   Meadow2.70 107 (3.72 106)2955 (345)64 (9)0.89–0.957.94 104 (9.69 103)814 (67)172 (17)0.83–0.96
   Transitional1.03 105 (9.93 103)1367 (132)148 (21)0.86–0.974.27 104 (4.02 103)889 (99)167 (26)0.85–0.94
   Fen2.96 107 (4.27 106)3278 (423)53 (8)0.90–0.961.35 104 (1.67 103)710 (103)174 (31)0.80–0.90
   Wet fen1.08 106 (1.63 105)2162 (301)111 (13)0.88–0.972.31 104 (3.20 103)950 (169)162 (19)0.77–0.91
Table 4. Model Parameters of the Potential Enzyme Activities Using the Flexible Temperature Dependence Function (Equation (2))a
 Flexible Model [Lloyd and Taylor, 1994], f(T) = A1 · image
ß-GlucosidaseXylosidaseN-Acetyl-ß-Glucosaminidase
A1B1T0A1B1T0A1B1T0
  • a

    Units are A1: (nmol MUB g−1 min−1) (ß-glucosidase, ß-xylosidase, N-acetyl-ß-glucosaminidase), (nmol AMC g−1 min−1) (tyrosine amino-peptidase, leucine amino-peptidase) and B1, T0: (K). Values are means (± standard errors).

Summer         
   Meadow2.68 104 (2.63 103)814 (102)204 (32)4.44 105 (5.36 104)1291 (126)179(24)4.71 104 (5.73104)1108 (131)179 (25)
   Transitional6.07 104 (5.73 103)990 (103)195 (25)2.17 105 (2.35 104)1159 (134)182 (16)9.31 104 (1.03 105)1203 (115)176 (16)
   Fen4.36 104 (3.19 103)812 (162)199 (32)2.12 106 (3.83 105)1684 (212)163 (23)1.13 104 (1.52 104)988 (135)184 (25)
   Wet fen5.93 103 (7.98 102)605 (76)209 (19)5.43 104 (4.84 103)833 (116)197 (24)5.76 106 (7.20 105)1776 (215)155 (21)
Autumn         
   Meadow1.41 104 (6.97 103)673 (59)212 (32)8.86 105 (3.91 105)1482 (184)174 (28)2.27 106 (8.32 105)1578 (199)168 (25)
   Transitional9.83 103 (5.51 102)581 (78)215 (35)3.25 105 (6.23 104)1250 (134)178 (26)7.23 106 (2.74 106)1830 (204)159 (20)
   Fen2.14 105 (1.94 104)1019 (121)195 (38)1.96 106 (3.73 105)1596 (301)169 (29)4.65 106 (2.93 106)1992 (267)148 (26)
   Wet fen3.26 104 (3.73 103)786 (98)206 (36)3.83 105 (9.21 104)1132 (125)186 (31)6.63 105 (6.34 105)1627 (165)161 (24)
Winter         
   Meadow1.27 104 (1.52 103)638 (71)211 (39)2.17 105 (3.21 104)1073 (96)187 (19)8.68 106 (2.23 105)1489 (168)168 (20)
   Transitional1.03 104 (3.45 103)674 (78)206 (32)5.51 104 (7.23 104)1371 (165)171 (28)2.69 105 (9.34 104)1298 (168)179 (26)
   Fen7.34 104 (7.24 103)1005 (124)195 (26)3.23 105 (5.32104)1194 (187)184 (36)2.65 106 (7.31 105)1783 (362)158 (33)
   Wet fen6.54 103 (6.24 102)590 (56)214 (25)1.27 106 (1.94105)1475 (243)173 (28)1.93 105 (5.34 104)1273 (154)171 (25)
 
 Tyrosine Amino-PeptidaseLeucine Amino-Peptidase   
 A1B1T0A1B1T0   
Summer         
   Meadow8.34 106 (1.03 105)2950 (523)86 (10)7.54 106 (5.2 105)2965 (368)81 (13)   
   Transitional9.62 106 (2.39 105)2765 (341)98 (13)3.44 107 (6.43 106)3083 (403)91 (9)   
   Fen7.58 106 (1.74 105)2703 (503)103 (21)4.87 106 (8.21 105)2960 (38999 (16)   
   Wet fen7.11 107 (9.21 106)3359 (322)83 (11)2.49 106 (9.21 105)2241 (421)113 (22)   
Autumn         
   Meadow4.07 106 (4.72 105)2292 (298)121 (19)7.37 105 (5.32 104)1770 (268)139 (16)   
   Transitional9.42 106 (3.53105)2584 (268)106 (14)9.12 104 (8.11 103)1380 (223)151 (27)   
   Fen6.95 108 (4.55 107)3896 (532)83 (15)6.52 106 (9.21 105)2410 (435)114 (30)   
   Wet fen2.46 104 (3.93 103)1664 (265)157 (28)3.29 106 (7.30 105)2678 (343)131 (22)   
Winter         
   Meadow1.12 105 (1.45 104)1437 (165)153 (15)7.17 105 (5.23 104)2119 (408)109 (17)   
   Transitional5.26 106 (3.61 105)2416 (321)113 (12)2.09 107 (5.12106)2930 (355)92 (13)   
   Fen1.12 104 (1.54 103)1016 (186)179 (17)8.11 105 (7.87 104)1674 (304)147 (21)   
   Wet fen6.15 104 (6.77103)1367 (156)146 (19)5.06 106 (4.48 105)2193 (412)123 (14)   

[22] The ratio of RTS(0–5)/RTS(13–18) was used to demonstrate differences of the relative temperature dependence between microbial activities, sampling dates and study sites independent of different activation energies of the microbial activities (caused by the different applied methods/or artificial substrates) (Figure 4). There were no statistical differences of the RTS(0–5)/RTS(13–18) of any studied activity (C and N mineralization, GLUC, XYL, NAG, TYRO, LEU) for sampling date (3.0 ≥ Fdate ≥ 1.1, 0.061 ≥ p ≤ 0.868), study site (2.0 ≥ Fsite ≥ 0.1, 0.120 ≥ p ≤ 0.977) and sampling date × study site (0.9 ≥ Fdate × Fsite ≥ 0.2, 0.484 ≥ p ≤ 0.975) (two-way ANOVA).

Figure 4.

Ratio of the relative temperature sensitivity (RTS(0–5) and RTS(13–18)) for all soil activities (study site: alpine meadow); C mineralization (Cmin), N mineralization (Nmin), and enzyme activities: Tyrosine amino-peptidase (TYRO), leucine amino-peptidase (LEU), ß-glucosidase (GLUC), ß-xylosidase (XYL), and N-acetyl-ß-glucosaminidase (NAG). Bars represent the mean, and whiskers indicate standard error.

[23] The mixed three-factorial model was used to identify differences between the RTS(0–5)/RTS(13–18) of the studied activities. Differences were significant between the activities (Factivity = 9.2, p ≤ 0.001). The ratios (all sampling dates and study sites) were, in ascending order: C mineralization (1.08), TYRO (1.11), LEU (1.13), N mineralization (1.14), XYL (1.20), NAG (1.22) and GLUC (1.25). The factor sampling date was significant (Fdate = 6.6, p ≤ 0.002), with the overall lowest RTS(0–5)/RTS(13–18) value (all study sites and activities) in autumn (1.12), followed by summer (1.16) and winter (1.20). The study site was not important (Fsite = 1.4, p ≤ 0.231) and all factor interactions were also not significant (0.7 ≥ F ≥ 0.0; 0.755 ≥ p ≤ 0.100).

4. Discussion

4.1. Temperature Sensitivity of C and N Mineralization

4.1.1. Constant Versus Flexible Temperature Response

[24] The temperature dependence of activities may vary depending on the investigated temperature range. In our study, there was no significant increase in the relative temperature sensitivity of C and N mineralization with decreasing temperature at any study site and sampling date within the investigated temperature range (0 to 30°C). The predicting ability (R2) of both models (simple exponential equation and flexible model according to Lloyd and Taylor [1994]) was similar for a given data set. This result indicates that the temperature dependence of C and N mineralization is adequately explained by a constant temperature response. Mikan et al. [2002] found similar correlation coefficients of both models for six Arctic tundra soils, although the temperature range (+0.5 to 14°C) was considerably smaller than in our study. A constant relation of basal respiration to temperature was also reported for larger temperature ranges in alpine [Reichstein et al., 2000], Antarctic [Hopkins et al., 2006; Smith, 2003], temperate [Pietikäinen et al., 2005] and Mediterranean [Fierer et al., 2003] regions in recent studies. In contrast, a considerable increase of the relative temperature dependence with decreasing temperature for C mineralization was found by Kirschbaum [1995] and Lloyd and Taylor [1994]. Calculating the temperature response from published data of several decomposition studies, Kirschbaum [1995] found a temperature dependency about 3 times higher at 0°C than at 20°C. Using 15 data sets from field experiments, Lloyd and Taylor [1994] detected a relative temperature increase about 2 times higher at 0°C than at 20°C. In contrast, some studies have even reported an increasing temperature sensitivity of C mineralization with increasing temperature for cold-adapted Arctic soils [Nadelhoffer et al., 1991].

[25] The temperature sensitivity of net N mineralization has also been found to be constant over a wide temperatures range and different soils (Kladivko and Keeney [1987] (temperature range: 10 to 30°C) and Stanford et al. [1973] (temperature range: 5 to 35°C)). In contrast, Kirschbaum [1995] reported that net N mineralization and denitrification appeared to have an overall lower temperature response with decreasing temperature than C mineralization. However, he concluded that the high variability in the published data prohibits any final conclusion. For net N mineralization, a lower increase of the relative temperature sensitivity at lower temperature may be influenced by nitrification and/or denitrification losses; these probably have different temperature responses than ammonification. Since potential nitrification rates were below the detection limit, we suggest that nitrification and/or denitrification did not influence net N mineralization in our study. There is evidence, however, that the mineralization immobilization turnover (MIT) may bias the temperature dependency. Niklińska et al. [1999] found no relationship in the net N mineralization rate of seven European humus samples with incubation temperatures below 15°C compared to higher incubation temperatures. They concluded that, at lower temperatures, net N immobilization is more dominant than net N mineralization at lower temperatures than net N mineralization. In agricultural soils incorporated with green manure, gross N immobilization showed a higher temperature dependency than gross N mineralization using the 15N pool dilution technique [Andersen and Jensen, 2001]. Thus the relative temperature dependence of net N mineralization may depend on the MIT, suggesting an increasing temperature response at lower incubation temperatures. In accordance with the classical C to N concept, there was no net immobilization (at any incubation temperature) for our investigated soils (C to N ratios between 17 and 22% %−1). Our method, however, does not distinguish between N immobilization and gross N mineralization. On the basis of the constant temperature dependence we found over a wide temperature range, N immobilization may not be a major N flux during the short incubation period. This is also supported by Wang et al. [2001], who found lower N immobilization under waterlogged conditions than under aerobic conditions for 20 different soils.

4.1.2. Q10 Values of the C and N Mineralization

[26] Incubation experiments have been criticized as underestimating temperature sensitivity because the quality of soil organic matter pools changes during incubation [Kirschbaum, 1995]. Accordingly, Reichstein et al. [2000] suggested calculating Q10 values from rate constants derived from longer incubation periods rather than from instantaneously derived activities at different temperature intervals. There is not much evidence that respiration was affected by changes in soil organic matter quality during incubation: The respiration rate over 6 weeks (at 20°C) was fairly constant in a preliminary experiment. The incubation experiment for the temperature response of C mineralization was carried out by increasing incubation temperature stepwise. Temperature sensitivity did not decrease with increasing temperature, which can be also interpreted to reflect constant substrate quality throughout the incubation. A highly uniform organic matter pool is supported for cold-adapted organic Arctic soils [Weintraub and Schimel, 2003] and for humus layers from boreal regions [Niklińska et al., 1999]. Hence the available Corg pool seems to be relatively large and uniform in cold-adapted environments.

[27] The methodologies for calculating Q10 values can vary significantly, and direct comparisons require caution [Fierer et al., 2005]. Therefore we only compared absolute Q10 values derived by simple first-order exponential models (constant Q10 values). The average Q10 value of C mineralization for all study sites and all sampling dates was 2.0 (range 1.5–2.7). The range of Q10 values of soil respiration fit into the range of reported temperature sensitivities (Q10 = 1.2 to 3.9) from different alpine soils [Dutzler-Franz, 1981]. Schinner and Gstraunthaler [1981] found a low Q10 value (1.5; 5–15°C) for an alpine meadow site in Austria, whereas higher values were reported for subalpine organic layers (2.5) and soils (2.8) between 5 and 25°C in the Swiss Alps [Reichstein et al., 2000] and from eight alpine soils (>3000 m above sea level) of the north American continent (Q10 = 2.5 to 3.8) [Fierer et al., 2006]. Very high temperature dependencies (Q10: 4.6 to 9.4) were reported for Arctic tundra soils [Mikan et al., 2002], while Smith [2003] found a median Q10 value of 2.0 (range 1.8 to 2.5) for 100 sites representing 21 habitats on a sub-Antarctic island.

[28] The average Q10 values of the N mineralization under waterlogged conditions in this study were 1.7 (range 1.3–2.8) for all soils. Our results are in line with Stanford et al. [1973] and Kladivko and Keeney [1987], who found an average of 2.0 for a temperature range of 5 to 35°C (11 different soils) and 10 to 35°C (one soil). Both studies were carried out under aerobic conditions. There might be differences in the Q10 of aerobic and waterlogged conditions. However, Kladivko and Keeney [1987] found no water-temperature interaction using a wide range of water potentials (−0.1 to −3.0 bar). Thus there is no expectation that waterlogged conditions yield lower temperature sensitivities than aerobic incubations.

4.2. Seasonal Pattern in Temperature Sensitivity of Soil-Derived Activities

[29] Our data showed no distinct seasonal pattern. The Q10 values of C mineralization were slightly lower in summer than in autumn and winter. The Q10 values of N mineralization were highest in winter, but the differences between seasons were generally small. During the relatively long incubation time (25 d) of the C mineralization experiment, soils may have adapted to the corresponding incubation temperature. However, a 4-d incubation might reveal no evidence for an adaptation of N mineralization rates. The RTS increase with decreasing temperature was greatest in winter for all investigated activities (C and N mineralization and enzyme activities). Little information is available about the seasonal adaptation of the temperature response of soil activities. Fierer et al. [2006] found that 17% of the variability in Q10 values of C mineralization (77 soil samples) was positively correlated to the mean monthly soil temperature but not to the mean annual temperatures at the study sites - even though soils may have adapted to the surrounding temperature after being sampled and transported to laboratory. Similarly, a seasonal adaptation of potential respiration (substrate: potassium glutamate) occurred in an alpine meadow of the Rocky Mountains [Lipson et al., 2002]. Those authors found lower temperature sensitivity in winter than in summer. Also, the temperature response of oxygen consumption was seasonally affected in coastal sediment, being higher in summer than in winter [Thamdrup et al., 1998]. In contrast, Niklińska et al. [1999] found increasing Q10 values of respiration from humus samples as mean annual temperatures at the study sites dropped. The relatively low differences we found in the temperature response for the C and N mineralization over the year indicates that the soil activities are well adapted to enable nutrient cycling in this cold environment.

4.3. Effect of Substrate Quality on the Temperature Response of C and N Mineralization

[30] The Q10 values of C and N mineralization were inversely related to substrate quality indices (exponential constant A). Other studies also report an inverse relationship between Q10 values of C mineralization and the substrate quality index [e.g., Fierer et al., 2005; Fierer et al., 2006; Mikan et al., 2002]. This supports the substrate quality-to-temperature hypothesis, which implies higher activation energy for lower substrate quality. According to Bosatta and Ågren [1999], the carbon quality of the soil organic matter can be defined as the total number of enzymatic steps required to mineralize carbon to the end product CO2. Hence more enzymatic steps lead to higher activation energy for the C mineralization and greater temperature sensitivity. In contrast, no temperature-to-substrate quality relationship was found in mineral soils [Kätterer et al., 1998; Reichstein et al., 2005]. Soil mineral particles can stabilize organic C through adsorption to surfaces, potentially resulting in different temperature dependencies [Mikan et al., 2002]. Therefore organic soils may better reflect the temperature dependence of biological processes than mineral soils.

[31] The nitrogen quality of soil organic matter was defined following the same concept. The relationship (slope and y axis intercept of the regression line) between the substrate quality index and the Q10 were similar for C and N mineralization. This suggests a similar temperature sensitivity for both processes under equal substrate quality conditions. Assuming a generally constant C to N ratio in soils over longer time periods, we would expect a similar quality index for both activities. However, the index was higher for N than C mineralization, pointing to differences in the use of soil organic matter pools. For C mineralization, soils were allowed to equilibrate until the respiration rate was constant (after the hyperlabile substrate pool was respired). According to Boone [1990], N transformations measured under waterlogged conditions partly use easily degradable substrates from aerobic soil organisms killed by the anaerobic test conditions. This may help explain the higher substrate quality indices of N mineralization (AN) than C mineralization (AC) in our study.

[32] It has been shown that the temperature sensitivity depend on the substrate quality index rather than on any other soil chemical property or microbial biomass. Similarly, Fierer et al. [2006] found Q10 values of C mineralization to be unrelated to soil chemical properties or microbial biomass C. Accepting that substrate quality is the key factor in the temperature sensitivity of soil-derived processes, then the turnover of more recalcitrant components should more strongly depend on temperature. In the framework of the global warming issue, this could alter the composition of the soil organic matter in peat soils.

4.4. Relative Temperature Sensitivity (RTS) of the Enzyme Activities

[33] Extracellular enzymes are important because they catalyze the rate-limiting steps of decomposition and nutrient cycling [Sinsabaugh, 1994]. Our study revealed significant differences in the relative temperature sensitivity (RTS) between amino-peptidases (LEU, TYRO) and enzymes involved in the C cycle (GLUC, XYL, NAG). The greater increase of the RTS of GLUC, XYL and NAG at lower temperature compared to the amino-peptidase (TYRO, LEU) and C and N mineralization suggest that temperature is an important factor regulating the use of different substrates. Hopkins et al. [2006] found a much higher temperature response for glucose-activated respiration at the lower temperature range (Q10 = 3.3 to 6.9; temperature range: −0.5 to 20°C) versus the upper temperature range (Q10 = 2.0 to 3.6; temperature range: 9 to 20°C) for five different Antarctic soils. In contrast, the temperature sensitivity of C mineralization of these soils was constant over the entire temperature range, having lower Q10 values (1.2 to 3.3). These results suggest that the turnover of easily degradable C substrates (like glucose) is more sensitive to temperature than higher molecular compounds, at least for cold soils. The almost constant RTS for TYRO and LEU in our study suggests that polypeptide decay is more favoured at lower temperatures. Similarly, N mineralization is disproportionally high (versus C mineralization) in easily degradable green manures at low temperatures in cool temperate agroecosystems [Magid et al., 2001]. However, soils from different climate regions may behave differently. Trasar-Cepeda et al. [2007] found much smaller RTS differences for a range of investigated enzymes of the C, N and S cycles in Mediterranean soils in NW Spain. In our study, the disproportional RTS pattern for different enzymes of the N and C cycle may be characteristic for alpine ecosystems and allow a sufficient supply of nitrogen during the cold season in these N-limited cold environments.

5. Conclusion

[34] This study revealed a temperature sensitivity of C and N mineralization in the lower range of reported values from the literature. The relatively low temperature response may be an adaptation to a cold environment in order to maintain nutrient cycling during the cold winter periods. No distinct seasonal adaptation was evident, but the within- and among-site variability in the temperature response was inversely related to substrate quality. However, the wide range of temperature responses for C and N mineralization reported in the literature cannot be explained by substrate quality alone. This calls for future investigations to better understand the temperature sensitivity of soil-derived processes. The degradation of easily available C polymers was more strongly inhibited by low temperature than the decay of peptides by peptidases. Moreover, a surplus of easily available N compounds versus C compounds at lower temperature may be important for ecosystem functioning in cold-adapted and N-limited environments. Our results underline the key role of temperature in regulating the substrate use from different fractions of soil organic matter.

Acknowledgments

[35] This research was funded by the DFG project 768. We would like to thank Meinhard Strobel (alpine research station, University of Innsbruck) for his hospitality during soil sampling. Additionally, we thank Andreas Büchse (University of Hohenheim) for helpful hints about the statistics used in this study.

Ancillary