| 英文摘要: | The large uncertainty in soil carbon–climate feedback predictions has been attributed to the incorrect parameterization of decomposition temperature sensitivity (Q10; ref. 1) and microbial carbon use efficiency2. Empirical experiments have found that these parameters vary spatiotemporally3, 4, 5, 6, but such variability is not included in current ecosystem models7, 8, 9, 10, 11, 12, 13. Here we use a thermodynamically based decomposition model to test the hypothesis that this observed variability arises from interactions between temperature, microbial biogeochemistry, and mineral surface sorptive reactions. We show that because mineral surfaces interact with substrates, enzymes and microbes, both Q10 and microbial carbon use efficiency are hysteretic (so that neither can be represented by a single static function) and the conventional labile and recalcitrant substrate characterization with static temperature sensitivity is flawed. In a 4-K temperature perturbation experiment, our fully dynamic model predicted more variable but weaker soil carbon–climate feedbacks than did the static Q10 and static carbon use efficiency model when forced with yearly, daily and hourly variable temperatures. These results imply that current Earth system models probably overestimate the response of soil carbon stocks to global warming. Future ecosystem models should therefore consider the dynamic interactions between sorptive mineral surfaces, substrates and microbial processes. Most ecosystem models used for soil carbon–climate feedback predictions use the turnover pool based structure and static Q10 for soil carbon dynamics7, 8, but these models underestimate soil carbon variability14 and predict very uncertain soil carbon stocks15. Some recent microbe-explicit models, aiming to improve soil carbon modelling, explicitly consider microbe–mineral–surface interactions9, 10, 11, 12, 13. These models have shown that microbial carbon use efficiency (CUE) is an important controller of carbon decomposition in response to temperature change11, 12, but dynamic interaction of CUE with temperature-dependent adsorption is rarely investigated (except see ref. 9). Further, in representing respiration and its response to temperature change, many of these microbe-explicit models impose static CUE (refs 9, 10, 11, 12), and some even characterize carbon substrates using the conventional ‘labile’ and ‘recalcitrant’ paradigm13, but empirical experiments5, 6, 16 and our results described below challenge each of these concepts. In addition to binding to polymeric soil organic matter (SOM), extracellular enzymes can adsorb to mineral surfaces and temporarily lose their capacity to degrade SOM (ref. 17). Our model (Supplementary Fig. 1) therefore allows SOM and mineral surfaces to compete for extracellular enzyme binding, such that increasing mineral surface area inhibits SOM degradation into dissolvable organic matter (DOM), all else equal. Simultaneously, DOM competes with extracellular enzymes for mineral surface adsorption and mineral surface adsorption competes with microbes for DOM. The model forms a network of SOM, DOM, microbes, extracellular enzymes and mineral surfaces, and models their competitive interactions using equilibrium chemistry approximation kinetics18. We predicted CUE using the dynamic energy budget (DEB) theory19, which allows for a thermodynamically consistent treatment of the balance between structural maintenance, structural growth and extracellular enzyme production in microbial metabolism. Our DEB model includes an internal reserve pool, which buffers between environmental substrate uptake and microbial cell metabolism. A reserve pool could increase microbes’ plasticity under environmental stress20. We illustrate the role of microbial plasticity by analysing a second model, identical except that it has no reserve pool (called a ‘rigid’ microbe). To resolve the variability of the soil carbon decomposition temperature sensitivity, in contrast to using a static Q10 (or Arrhenius activation energy) and CUE, we explicitly modelled the temperature dependencies (Methods) of enzymatic SOM degradation, microbial DOM uptake, microbial reserve pool turnover, mineral surface sorption and microbial maintenance, and implicitly for microbial cell growth and enzyme production (see Supplementary Methods). We calibrated (Methods) and evaluated the model (Supplementary Table 1) to be qualitatively consistent with 14 emergent empirical metrics (Supplementary Table 2) and addressed parameterization uncertainty through perturbation simulations. We identified three salient emergent responses from our transient simulations (Fig. 1). First, higher mineral surface adsorption capacity leads to lower respiration per total soil carbon mass (see contrast between Fig. 1a–c). Second, temperature sensitivity has large variability, depending, to various degrees, on many properties of the system. Third, the daily averaged respiration (red and green solid lines in Fig. 1) has lower temperature sensitivity and smaller range than does the original hourly respiration (blue and grey dots in Fig. 1), implying that models derived from coarse temporal resolution (daily) data will lead to error when applied at fine temporal resolution (hourly scales).
| http://www.nature.com/nclimate/journal/v5/n1/full/nclimate2438.html
|
