• Open Access

Thermodynamic theory explains the temperature optima of soil microbial processes and high Q10 values at low temperatures

Authors


Abstract

Our current understanding of the temperature response of biological processes in soil is based on the Arrhenius equation. This predicts an exponential increase in rate as temperature rises, whereas in the laboratory and in the field, there is always a clearly identifiable temperature optimum for all microbial processes. In the laboratory, this has been explained by denaturation of enzymes at higher temperatures, and in the field, the availability of substrates and water is often cited as critical factors. Recently, we have shown that temperature optima for enzymes and microbial growth occur in the absence of denaturation and that this is a consequence of the unusual heat capacity changes associated with enzymes. We have called this macromolecular rate theory – MMRT (Hobbs et al., 2013, ACS Chem. Biol. 8:2388). Here, we apply MMRT to a wide range of literature data on the response of soil microbial processes to temperature with a focus on respiration but also including different soil enzyme activities, nitrogen and methane cycling. Our theory agrees closely with a wide range of experimental data and predicts temperature optima for these microbial processes. MMRT also predicted high relative temperature sensitivity (as assessed by Q10 calculations) at low temperatures and that Q10 declined as temperature increases in agreement with data synthesis from the literature. Declining Q10 and temperature optima in soils are coherently explained by MMRT which is based on thermodynamics and heat capacity changes for enzyme-catalysed rates. MMRT also provides a new perspective, and makes new predictions, regarding the absolute temperature sensitivity of ecosystems – a fundamental component of models for climate change.

Introduction

The response of soil biochemical processes to changing temperature (T) is critical for predicting daily, seasonal and annual variation of biological cycling of nutrients and carbon (C). Soil biochemical processes are strongly temperature dependent in two broad ways: (i) Intrinsic – the direct effect of temperature on microbial metabolism which is largely driven by enzymatic kinetics; and (ii) Extrinsic – indirect effects, such as through temperature controls on substrate solubility and diffusion, and freezing and thawing (Davidson & Janssens, 2006; Conant et al., 2011). Extrinsic effects are highly context specific in that they vary between soils and climates, whereas intrinsic effects are directly dependent on enzyme-catalysed biochemical reactions. While extrinsic effects are extremely important, the focus of the work presented here is primarily on intrinsic effects.

A wide range of different mathematical models of biochemical responses to changing temperature in soils have been proposed (Lloyd & Taylor, 1994; Kirschbaum, 2000; Davidson & Janssens, 2006; Tuomi et al., 2008; Sierra, 2012). Many of these models are based on the Arrhenius function, but this function does not predict an optimum temperature (Topt) above which the rate declines (Fig. 1a). The decline in the rate of activity in biological systems above Topt has been attributed to enzyme denaturation at higher temperatures coupled with complex regulatory temperature responses in the cell. Corkrey et al. (2012) modelled these effects on the basis of a single ‘master enzyme’ where the complex negative effects above Topt are rolled into an enzyme denaturation term. However, enzyme denaturation is not a coherent explanation in soil systems as enzyme denaturation generally occurs at higher temperatures than those that are commonly observed in soils. For example, the well-studied soil bacterium Bacillus subtilis has an optimal growth temperature of 39 °C and a typical enzyme from Bacillus subtilis has an unfolding (denaturation) mid-point temperature of 59 °C (Ruller et al., 2008).

Figure 1.

(a) The relationship between reaction rate and temperature for the Arrhenius equation, Lloyd and Taylor's equation, and the MMRT equation (at two values selected for ∆CP). (b) The derived Q10 values from each of the equations with increasing temperature. (c) The same functions plotted as the natural log of the rate vs. temperature.

A second poorly explained observation is that the relative temperature sensitivity of ecosystem biological reactions is greatest at low temperatures and declines as temperature increases. Relative temperature sensitivity has frequently been calculated as the Q10 for biological processes; Q10 is the ratio of rates determined 10 °C apart (Conant et al., 2011; Sierra, 2012). Empirical measurements show high values of Q10 at low temperatures and declining Q10 values as temperature increases (e.g. for respiration (Del Grosso et al., 2005; Hamdi et al., 2013; Kirschbaum, 1995)). Despite recognizing that relative temperature sensitivity is not constant with increasing temperature, a number of global climate models simply fix Q10 at 2 or 1.5 or assign fixed Q10 for different C pools (summarized in Friedlingstein et al. (2006)). While this may be a reasonable first approximation for relative temperature sensitivity, a theoretical basis for changing Q10 is needed to place models on a stronger footing. The development of any underpinning theory would need to account for changing relative temperature sensitivity and particularly for declines in Q10 from a mean value of ~4.8 at 0 °C to ~2 above 25 °C as recently reported by Hamdi et al. (2013) in a synthesis of 68 studies of respiration.

We have recently developed a thermodynamic theory of the temperature dependence of enzyme reaction rates and microbial growth rates (Hobbs et al., 2013). We have called this ‘Macromolecular Rate Theory (MMRT)’ as it takes into account critical thermodynamic properties of biological macromolecules and specifically those of enzymes. At the heart of the theory is the fact that the activation energy (EA from the Arrhenius equation) of enzyme-catalysed reactions is theoretically temperature dependent. The temperature dependence of EA arises from a large, negative change in heat capacity associated with enzyme-catalysed reactions. This change in heat capacity results in negative curvature in a temperature vs. rate plot and directly predicts a Topt for enzymatically driven processes in direct contrast to the standard Arrhenius function (Fig. 1a). Furthermore, MMRT accounts for the decline in reaction rates above Topt without the need to invoke enzyme denaturation at modest temperatures. This theory also specifically predicts high Q10 values at low temperatures for enzymes (Fig. 1b and c).

Here, our objective was to apply MMRT to soil biochemical data, to test whether this approach could sensibly explain observed temperature responses of soil processes including Topt and the decline in Q10 with increasing temperature. First, we summarize the theoretical basis of MMRT. Second, we fit the derived equations to the observed temperature response curves of individual soil processes from studies drawn from the literature and compare derived physical parameters to those known from enzyme studies (Hobbs et al., 2013). Lastly, we use this framework to demonstrate that the theory predicts that Q10 will decline with increasing temperature, comparing the resulting curve with soil respiration data sets collated by Hamdi et al. (2013).

Materials and methods

Theory

Temperature response and temperature optimum

Arrhenius first proposed how the chemical reaction rate varies with temperature in the 19th century. The Arrhenius function is: inline image where k is the rate constant, A is a pre-exponential factor, EA is the activation energy and R is the universal gas constant. Many contemporary descriptions of the temperature dependence of soil microbial processes are based on this function (Lloyd & Taylor, 1994; Kirschbaum, 1995, 2000). The concept enshrined by this function is the activation energy (EA) which is the energy barrier over which reactants must ‘jump’ to make the transformation into products. This transformation generally involves bond breaking and/or bond making. If the barrier to bond breaking or bond making is high, the rate will be slow, and conversely, if the barrier is low, the rate will be high. Today, this transformation is described as progressing through a ‘transition state’. The exponential relationship between rate and EA means that small changes in EA result in large changes in rate. Temperature is also included in the exponential term and thus, linear increases in temperature lead to exponential increases in rate.

The late 19th century also saw the development of the formal equations of thermodynamics by Gibbs, Maxwell and Boltzmann, where activation energy is substituted in the Arrhenius equation (EA) by the change in Gibbs free energy, ∆G. The superscript † denotes the transition state and thus ∆G denotes the difference in free energy between the reactants and the transition state. In turn, the Gibbs free energy can be calculated from the difference between the change in enthalpy for the reaction (∆H) and the change in entropy (∆S), so that ∆G = ∆H − TS.

Early in the 20th century, Eyring and Polyani developed Arrhenius' theory further and quantified the pre-exponential term, A, from the Arrhenius' equation: A = κkBT/h, where kB is Boltzmann's constant, h is Planck's constant and κ accounts for quantum mechanical effects and ‘recrossing’ of the transition state barrier. For simplicity, here we assume that κ is equal to 1. This new formalism was dubbed ‘transition state theory’ and follows directly from the parental Arrhenius equation.

display math(1)

This equation can be simplified by taking the natural log of both sides:

display math(2)

This equation has been used to describe the temperature dependence of a large number of different chemical reactions with remarkable accuracy across a very broad temperature range. One of the central assumptions of transition state theory is that ∆G, the activation barrier, is independent of temperature and for reactions involving small molecules in common solvents such as water, this is generally true. However, biological reactions are almost always mediated by macromolecules such as enzymes and these very large molecules are peculiar insofar as they have large heat capacities (CP) (Cooper, 2005). Heat capacity (CP) is an easily measured property of a system and is simply defined as the temperature dependence of the enthalpy (H) and entropy (S) for the system (and therefore, the Gibbs Free Energy, G, for the system). During catalysed reactions, there are large and significant changes in heat capacity between different states of enzymes and particularly the difference in heat capacity between the enzyme bound to the substrate for the reaction, and the enzyme bound to the transition state for the reaction (∆CP). Large values of ∆CP lead to a marked temperature dependence of ∆G (Oliveberg et al., 1995) as follows:

display math(3)

Thus, the activation enthalpy, ∆H, varies linearly with temperature, while the activation entropy, ∆S, varies with the natural log of the temperature. The extent of this temperature dependence is determined by ∆CP. Large negative values of ∆CP will lead to a very significant temperature dependence of ∆G and significant deviations from Arrhenius behaviour (Fig. 1). Conversely, if ∆CP is zero, ∆G will be independent of temperature and rates will follow the Arrhenius equation.

Combining Eqns (2) and (3) gives:

display math(4)

We have shown previously that this equation describes the temperature dependence of enzyme-catalysed rates with large, negative values for ∆CP (Hobbs et al., 2013). We have also shown that if we make minor mutations to an enzyme that change ∆CP for the reaction, we can significantly change the temperature dependence of the enzyme-catalysed rate (Hobbs et al., 2013). These results are all in the absence of any enzyme denaturation. Eqn (4) also scales to describe the temperature dependence of microbial growth rates with observed values of ∆CP that are commensurate with the values that we determine for enzymes (Hobbs et al., 2013).

Temperature sensitivity

A second area of interest is predicting the sensitivity of biological processes to increasing temperature often through the use of Q10. If we take the Arrhenius function (substituting ∆G for EA), we get Q10:

display math(5)

For values of ∆G and T that are typical of biological systems, this function predicts Q10 to lie between 2 and 3 and to be largely invariant with temperature (Fig. 1b, blue line).

However, deriving Q10 from Eqn (4) (i.e. including the influence of ∆CP), we arrive at:

display math(6)

where ∆H is linearly temperature dependent [i.e. H = HTo + ∆Cp (TTo)]. This Q10 function curves strongly with temperature and predicts higher values of Q10 at lower temperatures (Fig. 1b).

Application to soil biochemical processes

We have previously used Eqn (4) to successfully explain the temperature optimum of enzyme catalysis and microbial growth without the need for denaturation (Hobbs et al., 2013). However, any one soil sample will contain a consortium of microbial species each adapted to take advantage of slightly different temperature ranges and with their own Topt. We have shown that Topt and ∆CP are correlated (Hobbs et al., 2013) and so we can construct a simple model for a scenario of a consortium of microbial populations (Fig. 2). Each coloured curve in Fig. 2 represents a single microbial species with its own Topt and ∆CP. When these curves are summed (black line in Fig. 2), the resulting curve also has a single Topt and ∆CP and is well described by Eqn (4). If we fit Eqn (4) to a soils data set then we determine the ‘community Topt’ and the ‘community ∆CP’ for the system.

Figure 2.

A simple consortium model showing the individual temperature-rate curves for six different organisms with increasing Topt values (rainbow coloured curves, blue-red). The solid black curve is the summation of the six individual rates and the grey line shows a standard Arrhenius function.

To test the applicability of this equation to soil systems, we recovered data from the literature where studies were conducted on the temperature response of a range of soil enzymes and soil processes. The intention was not to collect a complete set of temperature responses, but rather to find important representative examples. We were particularly careful to not include studies where soil moisture may have limited microbial activity. Where data were presented as figures rather than a table of rates, DataThief III version 1.6 (www.datathief.org) was used to extract the values. Data were converted into temperature in Kelvin and the natural log of all rate data calculated. Fits to Eqn (4) and the Arrhenius equation were performed in GraphPad Prism 6.00. The Arrhenius equation was only fitted to data up to Topt otherwise fits would be very poor as would be expected. When fitting Eqn (4), the reference temperature (T0) was set to four degrees below the experimental Topt value for convenience. For curve fitting, further criteria were imposed: data sets required at least six data points covering a temperature range of at least 10 °C, fitting started where the gradient of ln(rate) vs. temperature was greater than zero and lastly, the curvature of the data needed to be sufficiently well defined so as to constrain the fitting error for ∆CP to be less than 100%. The majority of the data sets with >6 data points covering >10 °C showed curvature, and thus the latter constraint of well-defined curvature was not particularly demanding. For example, we identified 59 respiration data sets from the literature, and of these, 28 data sets contained >6 data points covering >10 °C. Only five (of the remaining 28) data sets then failed the final criterion of ill-defined ∆CP values after fitting of Eqn (4). An obvious criticism in the analysis of these data using Eqn (4) is that we have simply introduced four new variables (∆H, ∆S, T0 and ∆CP) and that this will provide a good fit to these data irrespective of any explanatory power. In this context, it is important to note that ∆H, ∆S and T0 from Eqn (4) are tightly correlated and thus are not independent variables. Therefore, although Eqn (4) appears to be more complex than the Lloyd and Taylor or Kirschbaum functions, all of these functions are similar in that they effectively contain just two independent variables.

Application to changing Q10 with increasing temperature

One of the most studied processes in soil is respiration, and a recent study by Hamdi et al. (2013) provides an excellent synthesis of respiration data and Q10 data for soils incubated at a range of temperatures. They provided strong empirical evidence that Q10 declined with increasing temperature as also shown empirically by others (Kirschbaum, 1995, 2000; Del Grosso et al., 2005). We retrieved respiration and temperature data from supplementary materials in Hamdi et al. (2013). Full details for the criteria for data selection are given in Hamdi et al., but they focused on laboratory measurements (to avoid effects of root respiration) and selected studies where moisture content was not considered to be limiting and carbon sources had not been added.

Results and discussion

In this study, we focus on intrinsic (direct) controls of temperature for soil biological processes, specifically accounting for the temperature dependence of the activation energy of enzyme-catalysed reactions. Extrinsic effects would layer on top of the theory described here and the size of these extrinsic effects will be context specific, depending on design of the experiment and environment in which the relationship is tested. In general terms, the thermodynamic controls of extrinsic effects should follow the same basic principles and so will not necessarily mask intrinsic effects.

Representation of temperature response including Topt

Macromolecular rate theory (MMRT), as previously applied to enzymes and microbial growth rates in culture (Hobbs et al., 2013), also successfully describes the temperature dependence of a variety of soil enzyme assays and soil processes, both in the laboratory and in the field (Fig. 3 shows examples, additional studies and fits given in Table S1). MMRT predicted a Topt without requiring enzyme denaturation and Eqn (4) provided a better fit to rates at higher and lower temperatures than the Arrhenius-like functions (Fig. 3). If curvature is absent from the data set then there is little or no difference between fits based on Arrhenius or on MMRT.

Figure 3.

The fit of the MMRT equation (solid line) to rates of soil processes vs. temperature data taken from the literature: (a) amidase, propionamide/Lester data set (Frankenberger & Tabatabai, 1980), (b) myrosinase, Wooster data set (Al-Turki & Dick, 2003), (c) Cellulase, Nicollet data set (Deng & Tabatabai, 1994), (d) nitrification, nitrogen source 1 data set (Russell et al., 2002), (e) nitrification, nitrogen source 2 data set (Russell et al., 2002), (f) denitrification (Fischer & Whalen, 2005), (g) methane production, Pavia 4–9 days data set (Yao & Conrad, 2000), (h) methane oxidation, B1 data set (Dunfield et al., 1993), (i) methane oxidation (Whalen et al., 1990), (j) respiration, arable data set (Chen et al., 2010), (k) respiration, moisture corrected data set (Kirschbaum, 2000), and (l) forest soil respiration (Davidson et al., 2006). ΔCp values are in kJ mol−1 K−1 (±SE). For comparison, the dashed line in each panel is a standard Arrhenius plot fitted to data up to the Topt.

The observed ∆CP values derived for soil enzymes, processes and respiration from fitting Eqn (4) ranged from −0.9 to −12.2 kJ mol−1 K−1, similar in range to values for individual enzymes measured in vitro (Hobbs et al., 2013), and as we would expect when considering the ‘master enzyme’ approach proposed by Corkrey et al. (2012). The value of ∆CP defines the degree of curvature of the rate-vs.-temperature function and thus the temperature sensitivity of the process being measured (Fig. 2). Large negative values of ∆CP describe processes with lower temperature optima which also have greater fluctuations in temperature sensitivities (slope of lines in Fig. 2). This is particularly important at low temperatures where Arrhenius-like functions (Lloyd & Taylor, 1994; Kirschbaum, 1995) significantly underestimate the relative temperature sensitivity (see below discussion on Q10).

We recognize that the addition of ∆CP into the Arrhenius equation adds an extra variable that would provide greater flexibility to improve the fit to a given data set. However, ∆CP has a physical meaning for enzymes. Enzymes achieve formidable rate enhancements for chemical reactions at ambient temperatures and pressures by reducing the activation energy for the reaction. They achieve this reduction in activation energy by binding to, and stabilizing, the transition state for the reaction (Fersht, 1998). This tight binding of the transition state also reduces the number of conformational states of the enzyme and lowers the heat capacity of the enzyme-transition state complex. Thus, the change in heat capacity during an enzyme-catalysed reaction is the difference in heat capacity between the enzyme-transition state complex and the enzyme-substrate complex. This change in heat capacity is generally large and negative, i.e. the enzyme-transition state complex is significantly more rigid (with fewer degrees of freedom) when compared to the enzyme-substrate complex. Excluding this change in heat capacity during enzyme-catalysed reactions results in a continuous and exponential increase in reaction rate as predicted by the Arrhenius and other derived equations. Including ∆CP is therefore not only theoretically required to describe temperature response of biological systems, but predicts a temperature optimum at a lower temperature than denaturation of enzymatic systems. Clearly, at even higher temperatures, denaturation will be a dominant factor in reducing activity. However, we expect that the environment would select for microbial populations with Topt values matched to local temperature variations to avoid denaturation and irreversible damage to biological systems.

Relative temperature sensitivity – declining Q10 with increasing temperature

An understanding of the temperature sensitivity of biological responses is critical for accurate modelling of soil processes. Several authors have noted empirically that Q10 (as a metric of relative temperature sensitivity) declines with increasing temperature with a particular focus on respiration (Kirschbaum, 1995; Del Grosso et al., 2005; Hamdi et al., 2013). As noted above, Eqn (6) predicts that for soil processes there would be higher Q10 values at lower temperatures (Fig. 1b). Furthermore, Q10 curves predicted from Eqn (6) matched values from the data sets of respiration (Fig. 4a–d). Much larger values of Q10 were predicted by MMRT at low temperatures than predicted by the Arrhenius equation. Indeed even at modest values for ∆Cp (−5.3 kJ mol−1 K−1), MMRT predicts much higher Q10s (~6.0 at temperatures of 5 °C) than the Arrhenius equation (Q10 of ~3.0). At low incubation temperatures, Hamdi et al. (2013) frequently reported Q10 values well above 5 and in 10 data sets, values were greater than 50 for temperatures between −5 and 3 °C. While empirical measurements of Q10 at near 0 °C temperatures are generally lacking or highly variable, MMRT would predict that these values could be as high as 20–100, which would have important implications for cold environments (such as the high latitude land of the northern hemisphere). This prediction needs to be tested by careful measurements of soil microbial processes at low (near 0 °C) temperatures.

Figure 4.

(a–d) Q10 values determined at different temperatures drawn from several of the data sets examined by Hamdi et al. (2013) and fits of the Arrhenius Q10 function (blue line) and the MMRT Q10 function [red line, Eqn (6)]. (e) and (f) fits of Eqn (6) to all data from Hamdi et al. (red line), data ≥4 °C (green line) and data ≥10 °C (purple line). Note differences in y-axis scale between panels (e) and (f) to allow greater visualization of the lower Q10 values.

A number of empirical variations on the Arrhenius equation have been previously proposed to account for the decline in relative temperature sensitivity with increasing temperature and temperature optima (Lloyd & Taylor, 1994; Kirschbaum, 2000; Del Grosso et al., 2005). While these semiempirical fits to experimental data are useful for modelling purposes, the theoretical framework proposed by MMRT (developed from first principles and thermodynamics) justifies including a decline in Q10 as temperature increases. The precise value and rate of decline in Q10 with temperature predicted by MMRT depends on the ∆Cp value used in Eqn (6) (e.g. Fig 4e and f shows fits of Eqn (6) to all data (red line), data ≥4 °C (green line) and data ≥10 °C (purple line)). The derived ∆Cp values are −9.3, −1.8 and −0.5 kJ mol−1 K−1 respectively. Mathematically, as Topt increases (as expected for warmer environments), ∆Cp becomes smaller and we would expect the biological response of ecosystems to have ∆Cp values that approach zero as temperature increases. This is illustrated in Fig. 4 e and f when we exclude data at low temperatures, the derived ∆Cp values approach zero. When ∆Cp is 0, then the Q10 is that predicted by the Arrhenius curve [Eqn (5)] Q10 ~ 2–3. To utilize MMRT within ecosystem models, we need to determine the value of ∆Cp for different temperature ranges (particularly low temperatures where ∆Cp may vary substantially) and determine whether a continuous function of a variable ∆Cp can be derived and validated for different ecosystem temperatures.

Hamdi et al. (2013) collected a wide range of Q10 data from the literature and many investigators use Q10 as a proxy for temperature sensitivity in ecosystems. Sierra (2012) has recently noted that the first derivative of the Arrhenius function (i.e. dk/dT, the change in rate with temperature) is a more appropriate metric for absolute temperature sensitivity. Sierra also noted that the Q10 and dk/dT functions make contrasting predictions regarding the temperature sensitivity across different temperature regimes (high, medium & low temperatures) and the quality of carbon substrates [the carbon-quality temperature hypothesis, reviewed in Conant et al. (2011)]. The first derivative of the MMRT function (dk/dT) will similarly describe changes in temperature sensitivity based on the thermodynamic properties of enzyme-catalysed processes. The first derivatives of the Arrhenius and MMRT functions (dk/dT) are shown in Fig. 5, in comparison to the Q10 function. There are clear differences between these three functions as temperature increases. The Arrhenius derivative predicts continuous and increasing absolute temperature sensitivity with increasing temperature, whereas MMRT shows that temperature sensitivity initially rises, but then falls to 0 when the temperature optimum is approached. The veracity of this difference in predicted temperature sensitivity needs to be thoroughly tested, but may provide an explanation for different degrees of temperature sensitivities reported between studies [e.g. Conant et al. (2011), (Sierra, 2012)].

Figure 5.

The first derivative (dk/dT, y-axis, left hand side) of the Arrhenius function compared to that of the MMRT function vs. temperature (x-axis). The first derivative describes the absolute temperature sensitivity at any particular temperature. Thus, an MMRT system with a Topt of 35 °C will show a maximum temperature sensitivity at 23 °C and will be insensitive to temperature at 35 °C (dk/dT = 0). In contrast, the basic Q10 function [Eqn (5)] is plotted on the y-axis, right hand side, and gently declines over the temperature range.

We previously showed that MMRT applies to temperature dependence of in vitro enzyme rates and microbial growth (Hobbs et al., 2013) and here have demonstrated MMRT applies to soil processes. As much of life processes are enzymatically driven, we anticipate similar temperature dependence will also be found in higher organisms and potentially at ecosystem scales. Indeed, a declining Q10 of respiration of plants with increasing temperature is well established (Tjoelker et al., 2001; Atkin et al., 2005) but without a complete explanation. Whether MMRT can explain the temperature response of all biological systems remains an open question. A key unknown is determining representative, ensemble ∆Cp values for a range of ecosystems with different climates. Obtaining these ∆Cp values will require very careful measurement of biochemical responses in the field over a wide temperature range to accurately define curvature. These measurements need to be made when there is an absence of other constraining factors, such as low water availability, that hide the temperature signal. One of the data sets used here was in situ respiration rates measured from forest soils and this shows distinct curvature (Fig. 3l) that was well represented by MMRT (Davidson et al., 2006).

This new description of temperature dependence of enzyme-catalysed rates (macromolecular rate theory, MMRT) based on thermodynamics (Hobbs et al., 2013) applied well to soil biological rate data. This theory models realistic temperature optima for biological processes and further provides underpinning theoretical support for a decline in Q10 with increasing temperature frequently observed in soil studies. We suggest variations of Eqns (4) and (6) may be useful for incorporation into models of soil biological processes such as respiration and nutrient cycling, but we need to carefully determine ∆Cp values for different ecosystems and this is the subject of ongoing work in our laboratories. In particular, we emphasize the value of MMRT for accounting for sensitivity of rates to small changes in temperature in low temperature ecosystems as global warming effects become pronounced. Whether MMRT can explain other temperature responses of soil biology [e.g. the temperature sensitivity of different organic matter pools (Bosatta & Agren, 1999)] or more widely explain temperature response of “higher order” organisms (such as plants) is worth exploring.

Acknowledgements

Landcare Research (Ecosystems Services - C09X0705) and University of Waikato for funding, SR is supported by a grant from the New Zealand Agricultural Greenhouse Gas Research Centre. Discussions and feedback from many colleagues has been very beneficial for developing and testing these ideas including Dave Campbell, Johan Six, Miko Kirschbaum, Steve Frolking, Bryan Stevenson, Paul Mudge and Jack Pronger. We thank Hamdi and colleagues for putting their supplementary data on line and Carlos Sierra and an anonymous reviewer for constructive reviews.

Ancillary

Advertisement