全球变化知识资源中心

Climate is traditionally thought to be the predominant control on decomposition rates at global and regional scales, with biotic factors controlling only local rates^{2, 4}. Biotic factors are divided into decomposer organisms, such as soil microbes, and the quality (for example, chemical composition) of the plant litter they decompose. Recent work suggests that litter quality may be more important than climate in controlling decomposition rates across biomes worldwide^{3, 11}, but the influence of decomposer organisms is still assumed limited across broad climate gradients¹². A core reason for this assumption is that climate is considered a primary control on the activity of decomposers. As such, across climate gradients, mean temperature and moisture availability are assumed to explain much of the variation in the activity of decomposer organisms and hence decomposition rates of organic matter. These climate–decomposition relationships are used to parameterize and evaluate Earth-system models¹³. It is therefore important to test the assumption that climate drives decomposer activities because proper understanding of these activities is needed to inform model projections such as carbon cycle–climate feedbacks^{1, 14}.

Climate–decomposition relationships are typically developed from regional to global studies that use the mean response of decomposition to climate and litter quality drivers^{2, 3, 4, 5}. There is growing awareness in other areas of global change science that using mean responses masks the fine-scale variation required to understand effects of environmental change^{6, 7}, although the importance of local-scale variation to ecological processes was acknowledged decades ago⁹. When fine-scale variation is considered to predict species responses, for example, local factors are of equal or more importance than climate factors in determining phenomena such as species distributions⁶. Local conditions also dictate the effects of broad-scale global change factors on ecosystem processes. For example, warming effects on ecosystem respiration rates, and tree growth responses to elevated atmospheric CO₂, depend on soil carbon and nitrogen availability, respectively^{15, 16}. Local-scale factors might therefore strongly mediate regional- to global-scale responses to environmental change. If so, it is conceivable that in broad-scale decomposition experiments the focus on mean decomposition rates^{2, 3, 5, 12, 17} ‘averages away’ the role of local-scale factors in determining decomposer activity and hence organic matter decomposition rates. Using local-level variation—rather than location-level means—in decomposition experiments is then necessary to verify the conventional wisdom that climate is a primary control on decomposition rates at broad spatial scales.

Broad-scale decomposition experiments generally focus on the breakdown of foliar litters, leaving dead wood decomposition a critical uncertainty in carbon-cycle models. Global stores of dead wood are substantive; estimated at 73 ± 6 Pg C (ref. 8). For living wood carbon to be transformed to a longer-term carbon store in soils, it first passes through the dead wood pool¹⁸. The turnover rate and fate (for example, CO₂ versus soil carbon) of dead wood therefore influences the carbon balance of forests under global change^{19, 20}. Dead wood also stores plant nutrients and is a hotspot for nitrogen accumulation²¹, making its decomposition dynamics a determinant of forest productivity. Local-scale processes do affect wood decomposition where, for example, rates of mass loss depend on the species of wood-rot fungi and hyphal grazing by soil invertebrates^{22, 23}. The influence of these local-scale variables on wood decomposition at regional scales, relative to climate, is uncertain given the paucity of regional-scale wood decomposition studies²⁴.

Here we investigate whether climate or other factors primarily control wood decomposition rates across a regional gradient in temperate forest. Regional decomposition experiments typically include few replicates of a single litter type per location and analyse the mean of these observations^{2, 3, 5, 12, 17}, precluding detection of the relative effect of local-scale controls. To overcome this limitation we had 32 observations for a standard wood substrate at each of five locations. We proposed (Hypothesis 1a) that if climate is a predominant control on decomposition then the decomposition rates of replicate wood blocks should cluster around the location-level means (Fig. 1a). Conversely, (Hypothesis 1b) if local-scale factors are the predominant control on decomposition rates then replicate values should broadly scatter around the location-level means (Fig. 1b). Fungi are the primary decomposers of dead wood²⁵ and so we estimated decomposer activity by measuring percentage fungal colonization of the wood blocks (Supplementary Methods), a metric equivalent to that used to estimate the functional role of mycorrhizal fungi²⁶. If Hypothesis 1a holds, climate should largely explain fungal colonization and wood decomposition. If Hypothesis 1b holds, then fungal colonization (but not climate) should primarily explain wood decomposition rates. By using fungal colonization we explicitly recognize that local-scale controls—such as fungal grazers, density of dead wood as a source of wood-rot fungi, and nitrogen availability—probably vary both within and across locations. Our study was not designed to identify specific local-scale controls, but rather to test whether they need to be identified to inform the development of models used to project decomposition rates under changing climate.

a, The classical conceptual model where climate is the predominant control. b, A conceptualization where local-scale factors that affect decomposer activity are instead the predominant control on decomposition rates. Decomposition is represented as mass loss of plant litter, and climate as mean annual temperature. Decomposition rates and climate variables are, however, represented using various expressions, including rate constants (k) and functions that integrate mean monthly temperature and precipitation data. The representation of these variables affects the form of the relationship (for example, linear versus curvilinear) but the relationships are always positive, as depicted above. The classical decomposition paradigm, shown in a, posits that climate explains (and controls) variation in decomposition rates at regional to global scales because climate functions as the primary control on the activity of decomposer organisms. In contrast, an emerging idea in projecting ecological responses to global change, shown in b, suggests instead that local-scale controls on biotic activity generate local-level variation in process rates equal to or greater than broad-scale controls such as climate, highlighting the need to understand local context-dependency to project decomposition rates under changing environmental conditions.

Full details are given in the Supplementary Methods. Briefly, our research was conducted at five locations spanning ~12° latitude in eastern US temperate, second-growth forest. At four sites in each location (20 sites total), 70-m transects were established and wooden blocks placed on the forest floor at 10-m intervals. Our design captured both local- and broad-scale spatial variation in climate by varying slope aspect (north- or south-facing), slope position (wetter downslope to drier upslope), and location (northern to southern sites). Although Coweeta is ~6° latitude from the most northern (Yale Myers) and southern (San Felasco) locations, the mountainous elevation makes it ecologically similar to Yale Myers. Wooden blocks (15 × 12 × 2 cm) were placed in the field for 13 months and consisted of untreated white pine. They were large enough to detect termite colonization and were modified to permit ant nesting by creating nesting chambers. Blocks were placed in contact with the surface soil and flush with the litter layer. Soil temperature and moisture were measured at each wood block on six occasions. Wood blocks were also checked for termite and ant colonization. Wooden blocks were retrieved, the fauna counted and weighed after drying at 65 °C, and then drilled to create sawdust for fungal biomass and total carbon determinations. Drilled wood blocks were re-weighed, dried at 65 °C to constant mass, and weighed again. We accounted for differences in initial masses across wood blocks by expressing mass loss as percentage carbon loss. Potential controls on decomposition were analysed with linear mixed models, permitting transect to be treated as a random effect and hence account for the spatial clustering of the experimental design (Supplementary Methods).

1. Wieder, W. R., Bonan, G. B. & Allison, S. D. Global soil carbon projections are improved by modelling microbial processes. Nature Clim. Change 3, 909–912 (2013).
   • CAS
   • ADS
   • Article
2. Berg, B. et al. Litter mass-loss rates in pine forests for Europe and Eastern United States: Some relationships with climate and litter quality. Biogeochemistry 20, 127–159 (1993).
   • Article
3. Currie, W. S. et al. Cross-biome transplants of plant litter show decomposition models extend to a broader climatic range but lose predictability at the decadal time scale. Glob. Change Biol. 16, 1744–1761 (2010).
   • Article
4. Meentemeyer, V. Macroclimate and lignin control of litter decomposition rates. Ecology 59, 465–472 (1978).
   • CAS
   • ISI
   • Article
5. Moore, T. R. et al. Litter decomposition rates in Canadian forests. Glob. Change Biol. 5, 75–82 (1999).
   • Article
6. Clark, J. S. et al. Individual-scale variation, species-scale differences: Inference needed to understand diversity. Ecol. Lett. 14, 1273–1287 (2011).
   • PubMed
   • Article
7. Mace, G. M. Ecology must evolve. Nature 503, 191–URL: