Response of soil surface CO2 flux in a boreal forest to ecosystem warming

Soil surface carbon dioxide (CO₂) flux (R_S) was measured for 2 years at the Boreal Soil and Air Warming Experiment site near Thompson, MB, Canada. The experimental design was a complete random block design that consisted of four replicate blocks, with each block containing a 15 m × 15 m control and heated plot. Black spruce [Picea mariana (Mill.) BSP] was the overstory species and Epilobium angustifolium was the dominant understory. Soil temperature was maintained (～5 °C) above the control soil temperature using electric cables inside water filled polyethylene tubing for each heated plot. Air inside a 7.3‐m‐diameter chamber, centered in the soil warming plot, contained approximately nine black spruce trees was heated ～5 °C above control ambient air temperature allowing for the testing of soil‐only warming and soil+air warming. Soil surface CO₂ flux (R_S) was positively correlated (P < 0.0001) to soil temperature at 10 cm depth. Soil surface CO₂ flux (R_S) was 24% greater in the soil‐only warming than the control in 2004, but was only 11% greater in 2005, while R_S in the soil+air warming treatments was 31% less than the control in 2004 and 23% less in 2005. Live fine root mass (< 2 mm diameter) was less in the heated than control treatments in 2004 and statistically less (P < 0.01) in 2005. Similar root mass between the two heated treatments suggests that different heating methods (soil‐only vs. soil+air warming) can affect the rate of decomposition.     

Controls on microbial CO2 production: a comparison of surface and subsurface soil horizons

Although a significant amount of the organic C stored in soil resides in subsurface horizons, the dynamics of subsurface C stores are not well understood. The objective of this study was to determine if changes in soil moisture, temperature, and nutrient levels have similar effects on the mineralization of surface (0–25 cm) and subsurface (below 25 cm) C stores. Samples were collected from a 2 m deep unsaturated mollisol profile located near Santa Barbara, CA, USA. In a series of experiments, we measured the influence of nutrient additions (N and P), soil temperature (10–35°C), and soil water potential (?0.5 to ?10 MPa) on the microbial mineralization of native soil organic C. Surface and subsurface soils were slightly different with respect to the effects of water potential on microbial CO₂ production; C mineralization rates in surface soils were more affected by conditions of moderate drought than rates in subsurface soils. With respect to the effects of soil temperature and nutrient levels on C mineralization rates, subsurface horizons were significantly more sensitive to increases in temperature or nutrient availability than surface horizons. The mean Q₁₀ value for C mineralization rates was 3.0 in surface horizons and 3.9 in subsurface horizons. The addition of either N or P had negligible effects on microbial CO₂ production in surface soil layers; in the subsurface horizons, the addition of either N or P increased CO₂ production by up to 450% relative to the control. The results of these experiments suggest that alterations of the soil environment may have different effects on CO₂ production through the profile and that the mineralization of subsurface C stores may be particularly susceptible to increases in temperature or nutrient inputs to soil.     

Experimental warming and burn severity alter soil CO2 flux and soil functional groups in a recently burned boreal forest

Global warming is projected to be greatest in northern regions, where forest fires are also increasing in frequency. Thus, interactions between fire and temperature on soil respiration at high latitudes should be considered in determining feedbacks to climate. We tested the hypothesis that experimental warming will augment soil CO₂ flux in a recently burned boreal forest by promoting microbial and root growth, but that this increase will be less apparent in more severely burned areas. We used open‐top chambers to raise temperatures 0.4–0.9°C across two levels of burn severity in a fire scar in Alaskan black spruce forest. After 3 consecutive years of warming, soil respiration was measured through a portable gas exchange system. Abundance of active microbes was determined by using Biolog EcoPlates^? for bacteria and ergosterol analysis for fungi. Elevated temperatures increased soil CO₂ flux by 20% and reduced root biomass, but had no effect on bacterial or fungal abundance or soil organic matter (SOM) content. Soil respiration, fungal abundance, SOM, and root biomass decreased with increasing burn severity. There were no significant interactions between temperature and burn severity with respect to any measurement. Higher soil respiration rates in the warmed plots may be because of higher metabolic activity of microbes or roots. All together, we found that postfire soils are a greater source of CO₂ to the atmosphere under elevated temperatures even in severely burned areas, suggesting that global warming may produce a positive feedback to atmospheric CO₂, even in young boreal ecosystems.     

CO2 flux from soil in pastures and forests in southwestern Amazonia

Stocks of carbon in Amazonian forest biomass and soils have received considerable research attention because of their potential as sources and sinks of atmospheric CO₂. Fluxes of CO₂ from soil to the atmosphere, on the other hand, have not been addressed comprehensively in regard to temporal and spatial variations and to land cover change, and have been measured directly only in a few locations in Amazonia. Considerable variation exists across the Amazon Basin in soil properties, climate, and management practices in forests and cattle pastures that might affect soil CO₂ fluxes. Here we report soil CO₂ fluxes from an area of rapid deforestation in the southwestern Amazonian state of Acre. Specifically we addressed (1) the seasonal variation of soil CO₂ fluxes, soil moisture, and soil temperature; (2) the effects of land cover (pastures, mature, and secondary forests) on these fluxes; (3) annual estimates of soil respiration; and (4) the relative contributions of grass‐derived and forest‐derived C as indicated by δ¹³CO₂. Fluxes were greatest during the wet season and declined during the dry season in all land covers. Soil respiration was significantly correlated with soil water‐filled pore space but not correlated with temperature. Annual fluxes were higher in pastures compared with mature and secondary forests, and some of the pastures also had higher soil C stocks. The δ¹³C of CO₂ respired in pasture soils showed that high respiration rates in pastures were derived almost entirely from grass root respiration and decomposition of grass residues. These results indicate that the pastures are very productive and that the larger flux of C cycling through pasture soils compared with forest soils is probably due to greater allocation of C belowground. Secondary forests had soil respiration rates similar to mature forests, and there was no correlation between soil respiration and either forest age or forest biomass. Hence, belowground allocation of C does not appear to be directly related to the stature of vegetation in this region. Variation in seasonal and annual rates of soil respiration of these forests and pastures is more indicative of flux of C through the soil rather than major net changes in ecosystem C stocks.     

Forest soil CO2 flux: uncovering the contribution and environmental responses of ectomycorrhizas

Forests play a critical role in the global carbon cycle, being considered an important and continuing carbon sink. However, the response of carbon sequestration in forests to global climate change remains a major uncertainty, with a particularly poor understanding of the origins and environmental responses of soil CO₂ efflux. For example, despite their large biomass, the contribution of ectomycorrhizal (EM) fungi to forest soil CO₂ efflux and responses to changes in environmental drivers has, to date, not been quantified in the field. Their activity is often simplistically included in the ‘autotrophic’ root respiration term. We set up a multiplexed continuous soil respiration measurement system in a young Lodgepole pine forest, using a mycorrhizal mesh collar design, to monitor the three main soil CO₂ efflux components: root, extraradical mycorrhizal hyphal, and soil heterotrophic respiration. Mycorrhizal hyphal respiration increased during the first month after collar insertion and thereafter remained remarkably stable. During autumn the soil CO₂ flux components could be divided into ∼60% soil heterotrophic, ∼25% EM hyphal, and ∼15% root fluxes. Thus the extraradical EM mycelium can contribute substantially more to soil CO₂ flux than do roots. While EM hyphal respiration responded strongly to reductions in soil moisture and appeared to be highly dependent on assimilate supply, it did not responded directly to changes in soil temperature. It was mainly the soil heterotrophic flux component that caused the commonly observed exponential relationship with temperature. Our results strongly suggest that accurate modelling of soil respiration, particularly in forest ecosystems, needs to explicitly consider the mycorrhizal mycelium and its dynamic response to specific environmental factors. Moreover, we propose that in forest ecosystems the mycorrhizal CO₂ flux component represents an overflow ‘CO₂ tap’ through which surplus plant carbon may be returned directly to the atmosphere, thus limiting expected carbon sequestration from trees under elevated CO₂.     

Soil CO2 efflux in a boreal pine forest under atmospheric CO2 enrichment and air warming

The response of forest soil CO₂ efflux to the elevation of two climatic factors, the atmospheric concentration of CO₂ (↑CO₂ of 700 μmol mol⁻¹) and air temperature (↑ T with average annual increase of 5°C), and their combination (↑CO₂+↑ T ) was investigated in a 4-year, full-factorial field experiment consisting of closed chambers built around 20-year-old Scots pines ( Pinus sylvestris L.) in the boreal zone of Finland. Mean soil CO₂ efflux in May–October increased with elevated CO₂ by 23–37%, with elevated temperature by 27–43%, and with the combined treatment by 35–59%. Temperature elevation was a significant factor in the combined 4-year efflux data, whereas the effect of elevated CO₂ was not as evident. Elevated temperature had the most pronounced impact early and late in the season, while the influence of elevated CO₂ alone was especially notable late in the season. Needle area was found to be a significant predictor of soil CO₂ efflux, particularly in August, a month of high root growth, thus supporting the assumption of a close link between whole-tree physiology and soil CO₂ emissions. The decrease in the temperature sensitivity of soil CO₂ efflux observed in the elevated temperature treatments in the second year nevertheless suggests the existence of soil response mechanisms that may be independent of the assimilating component of the forest ecosystem. In conclusion, elevated atmospheric CO₂ and air temperature consistently increased forest soil CO₂ efflux over the 4-year period, their combined effect being additive, with no apparent interaction.     

Source components and interannual variability of soil CO2 efflux under experimental warming and clipping in a grassland ecosystem

Partitioning soil CO₂ efflux into autotrophic (R_A) and heterotrophic (R_H) components is crucial for understanding their differential responses to climate change. We conducted a long‐term experiment (2000–2005) to investigate effects of warming 2°C and yearly clipping on soil CO₂ efflux and its components (i.e. R_A and R_H) in a tallgrass prairie ecosystem. Interannual variability of these fluxes was also examined. Deep collars (70 cm) were inserted into soil to measure R_H. R_A was quantified as the difference between soil CO₂ efflux and R_H. Warming treatment significantly stimulated soil CO₂ efflux and its components (i.e. R_A and R_H) in most years. In contrast, yearly clipping significantly reduced soil CO₂ efflux only in the last 2 years, although it decreased R_H in every year of the study. Temperature sensitivity (i.e. apparent Q₁₀ values) of soil CO₂ efflux was slightly lower under warming (P>0.05) and reduced considerably by clipping (P<0.05) compared with that in the control. On average over the 4 years, R_H accounted for approximately 65% of soil CO₂ efflux with a range from 58% to 73% in the four treatments. Over seasons, the contribution of R_H to soil CO₂ efflux reached a maximum in winter (∼90%) and a minimum in summer (∼35%). Annual soil CO₂ efflux did not vary substantially among years as precipitation did. The interannual variability of soil CO₂ efflux may be mainly caused by precipitation distribution and summer severe drought. Our results suggest that the effects of warming and yearly clipping on soil CO₂ efflux and its components did not result in significant changes in R_H or R_A contribution, and rainfall timing may be more important in determining interannual variability of soil CO₂ efflux than the amount of annual precipitation.     

The response of soil CO2 flux to changes in atmospheric CO2, nitrogen supply and plant diversity

We measured soil CO₂ flux over 19 sampling periods that spanned two growing seasons in a grassland Free Air Carbon dioxide Enrichment (FACE) experiment that factorially manipulated three major anthropogenic global changes: atmospheric carbon dioxide (CO₂) concentration, nitrogen (N) supply, and plant species richness. On average, over two growing seasons, elevated atmospheric CO₂ and N fertilization increased soil CO₂ flux by 0.57 µmol m^?2 s^?1 (13% increase) and 0.37 µmol m^?2 s^?1 (8% increase) above average control soil CO₂ flux, respectively. Decreases in planted diversity from 16 to 9, 4 and 1 species decreased soil CO₂ flux by 0.23, 0.41 and 1.09 µmol m^?2 s^?1 (5%, 8% and 21% decreases), respectively. There were no statistically significant pairwise interactions among the three treatments. During 19 sampling periods that spanned two growing seasons, elevated atmospheric CO₂ increased soil CO₂ flux most when soil moisture was low and soils were warm. Effects on soil CO₂ flux due to fertilization with N and decreases in diversity were greatest at the times of the year when soils were warm, although there were no significant correlations between these effects and soil moisture. Of the treatments, only the N and diversity treatments were correlated over time; neither were correlated with the CO₂ effect. Models of soil CO₂ flux will need to incorporate ecosystem CO₂ and N availability, as well as ecosystem plant diversity, and incorporate different environmental factors when determining the magnitude of the CO₂, N and diversity effects on soil CO₂ flux.     

The turnover of carbon pools contributing to soil CO2 and soil respiration in a temperate forest exposed to elevated CO2 concentration

Soil carbon is returned to the atmosphere through the process of soil respiration, which represents one of the largest fluxes in the terrestrial C cycle. The effects of climate change on the components of soil respiration can affect the sink or source capacity of ecosystems for atmospheric carbon, but no current techniques can unambiguously separate soil respiration into its components. Long‐term free air CO₂ enrichment (FACE) experiments provide a unique opportunity to study soil C dynamics because the CO₂ used for fumigation has a distinct isotopic signature and serves as a continuous label at the ecosystem level. We used the ¹³C tracer at the Duke Forest FACE site to follow the disappearance of C fixed before fumigation began in 1996 (pretreatment C) from soil CO₂ and soil‐respired CO₂, as an index of belowground C dynamics during the first 8 years of the experiment. The decay of pretreatment C as detected in the isotopic composition of soil‐respired CO₂ and soil CO₂ at 15, 30, 70, and 200 cm soil depth was best described by a model having one to three exponential pools within the soil system. The majority of soil‐respired CO₂ (71%) originated in soil C pools with a turnover time of about 35 days. About 55%, 50%, and 68% of soil CO₂ at 15, 30, and 70 cm, respectively, originated in soil pools with turnover times of less than 1 year. The rest of soil CO₂ and soil‐respired CO₂ originated in soil pools that turn over at decadal time scales. Our results suggest that a large fraction of the C returned to the atmosphere through soil respiration results from dynamic soil C pools that cannot be easily detected in traditionally defined soil organic matter standing stocks. Fast oxidation of labile C substrates may prevent increases in soil C accumulation in forests exposed to elevated [CO₂] and may consequently result in shorter ecosystem C residence times.     

Sensitivity of a dynamic global vegetation model to climate and atmospheric CO2

The equilibrium carbon storage capacity of the terrestrial biosphere has been investigated by running the Lund–Potsdam–Jena Dynamic Global Vegetation Model to equilibrium for a range of CO₂ concentrations and idealized climate states. Local climate is defined by the combination of an observation-based climatology and perturbation patterns derived from a 4 × CO₂ warming simulations, which are linearly scaled to global mean temperature deviations, Δ T _glob. Global carbon storage remains close to its optimum for Δ T _glob in the range of ±3°C in simulations with constant atmospheric CO₂. The magnitude of the carbon loss to the atmosphere per unit change in global average surface temperature shows a pronounced nonlinear threshold behavior. About twice as much carbon is lost per degree warming for Δ T _glob above 3°C than for present climate. Tropical, temperate, and boreal trees spread poleward with global warming. Vegetation dynamics govern the distribution of soil carbon storage and turnover in the climate space. For cold climate conditions, the global average decomposition rate of litter and soil decreases with warming, despite local increases in turnover rates. This result is not compatible with the assumption, commonly made in global box models, that soil turnover increases exponentially with global average surface temperature, over a wide temperature range.     

Vertical partitioning of CO2 production within a temperate forest soil

The major driving factors of soil CO₂ production – substrate supply, temperature, and water content – vary vertically within the soil profile, with the greatest temporal variations of these factors usually near the soil surface. Several studies have demonstrated that wetting and drying of the organic horizon contributes to temporal variation in summertime soil CO₂ efflux in forests, but this contribution is difficult to quantify. The objectives of this study were to partition CO₂ production vertically in a mixed hardwood stand of the Harvard Forest, Massachusetts, USA, and then to use that partitioning to evaluate how the relative contributions of CO₂ production by genetic soil horizon vary seasonally and interannually. We measured surface CO₂ efflux and vertical soil profiles of CO₂ concentration, temperature, water content, and soil physical characteristics. These data were applied to a model of effective diffusivity to estimate CO₂ flux at the top of each genetic soil horizon and the production within each horizon. A sensitivity analysis revealed sources of uncertainty when applying a diffusivity model to a rocky soil with large spatial heterogeneity, especially estimates of bulk density and volumetric water content and matching measurements of profiles and surface fluxes. We conservatively estimate that the O horizon contributed 40–48% of the total annual soil CO₂ efflux. Although the temperature sensitivity of CO₂ production varied across soil horizons, the partitioning of CO₂ production by horizon did not improve the overall prediction of surface CO₂ effluxes based on temperature functions. However, vertical partitioning revealed that water content covaried with CO₂ production only in the O horizon. Large interannual variations in estimates of O horizon CO₂ production indicate that this layer could be an important transient interannual source or sink of ecosystem C.     

A search for predictive understanding of plant responses to elevated [CO2]

This paper reviews two decades of effort by the scientific community in a search for predictive understanding of plant responses to elevated [CO₂]. To evaluate the progress of research in leaf photosynthesis, plant respiration, root nutrient uptake, and carbon partitioning, we divided scientific activities into four phases: (I) initial assessments derived from our existing knowledge base to provide frameworks for experimental studies; (II) experimental tests of the initial assessments; (III) in cases where assessments were invalidated, synthesis of experimental results to stimulate alternative hypotheses and further experimentation; and (IV) formation of new knowledge. This paper suggests that photosynthetic research may have gone through all four phases, considering that (a) variable responses of photosynthesis to [CO₂] are generally explainable, (b) extrapolation of leaf-level studies to the global scale has been examined, and (c) molecular studies are under way. Investigation of plant respiratory responses to [CO₂] has reached the third phase: experimental results have been accumulated, and mechanistic approaches are being developed to examine alternative hypotheses in search for new concepts and/or new quantitative frameworks to understand respiratory responses to elevated [CO₂]. The study of nutrient uptake kinetics is still in the second phase: experimental evidence has contradicted some of the initial assessments, and more experimental studies need to be designed before generalizations can be made. It is quite unfortunate that we have not made much progress in understanding mechanisms of carbon partitioning during the past two decades. This is due in part to the fact that some of the holistic theories, such as functional balance and optimality, have not evolved into testable hypotheses to guide experimental studies. This paper urges modelers to play an increasing role in plant–CO₂ research by disassembling these existing theories into hypotheses and urges experimentalists to design experiments to examine these holistic concepts.     

Responses of soil respiration to elevated CO2, air warming, and changing soil water availability in a model old-field grassland

Responses of soil respiration to atmospheric and climatic change will have profound impacts on ecosystem and global carbon (C) cycling in the future. This study was conducted to examine effects on soil respiration of the concurrent driving factors of elevated atmospheric CO₂ concentration, air warming, and changing precipitation in a constructed old‐field grassland in eastern Tennessee, USA. Model ecosystems of seven old‐field species were established in open‐top chambers and treated with factorial combinations of ambient or elevated (+300 ppm) CO₂ concentration, ambient or elevated (+3 °C) air temperature, and high or low soil moisture content. During the 19‐month experimental period from June 2003 to December 2004, higher CO₂ concentration and soil water availability significantly increased mean soil respiration by 35.8% and 15.7%, respectively. The effects of air warming on soil respiration varied seasonally from small reductions to significant increases to no response, and there was no significant main effect. In the wet side of elevated CO₂ chambers, air warming consistently caused increases in soil respiration, whereas in the other three combinations of CO₂ and water treatments, warming tended to decrease soil respiration over the growing season but increase it over the winter. There were no interactive effects on soil respiration among any two or three treatment factors irrespective of time period. Treatment‐induced changes in soil temperature and moisture together explained 49%, 44%, and 56% of the seasonal variations of soil respiration responses to elevated CO₂, air warming, and changing precipitation, respectively. Additional indirect effects of seasonal dynamics and responses of plant growth on C substrate supply were indicated. Given the importance of indirect effects of the forcing factors and plant community dynamics on soil temperature, moisture, and C substrate, soil respiration response to climatic warming should not be represented in models as a simple temperature response function, and a more mechanistic representation including vegetation dynamics and substrate supply is needed.     

The respiratory source of CO2

Abstract Approximately half of the carbon plants fix in photosynthesis is lost in dark respiration. The major pathways for dark respiration and their control are briefly discussed in the context of a growing plant. It is suggested that whole-plant respiration may be largely ADP-limited and that fine control of the respiratory network serves to select the respiratory substrate and to partition carbon between the numerous possible fates within the network. The striking stoichiometry between whole-plant growth and respiration is reviewed, and the relationships between substrate-limited growth and ADP-limited respiration are discussed.     

Precipitation pulses and soil CO2 flux in a Sonoran Desert ecosystem

Precipitation is a major driver of biological processes in arid and semiarid ecosystems. Soil biogeochemical processes in these water‐limited systems are closely linked to episodic rainfall events, and the relationship between microbial activity and the amount and timing of rainfall has implications for the whole‐system carbon (C) balance. Here, the influences of storm size and time between events on pulses of soil respiration were explored in an upper Sonoran Desert ecosystem. Immediately following experimental rewetting in the field, CO₂ efflux increased up to 30‐fold, but generally returned to background levels within 48 h. CO₂ production integrated over 48 h ranged from 2.5 to 19.3 g C m⁻² and was greater beneath shrubs than in interplant spaces. When water was applied on sequential days, postwetting losses of CO₂ were only half a large as initial fluxes, and the size of the second pulse increased with time between consecutive events. Soil respiration was more closely linked to the organic matter content of surface soils than to rainfall amount. Beneath shrubs, rates increased nonlinearly with storm size, reaching an asymptote at approximately 0.5 cm simulated storms. This nonlinear relationship stems from (1) resource limitation of microbial activity that is manifest at small time scales, and (2) greatly reduced process rates in deeper soil strata. Thus, beyond some threshold in storm size, increasing the duration or depth of soil moisture has little consequence for short‐term losses of CO₂. In addition, laboratory rewetting across a broad range in soil water content suggest that microbial activity and CO₂ efflux following rainfall may be further modified by the routing and redistribution of water along hillslopes. Finally, analysis of long‐term precipitation data suggests that half the monsoon storms in this system are sufficient to induce soil heterotrophic activity and C losses, but are not large enough to elicit autotrophic activity and C accrual by desert shrubs.     

Towards a better experimental basis for upscaling plant responses to elevated CO2 and climate warming

Few of the most common assumptions used in models of responses of plants and ecosystems to elevated CO₂ and climate warming have been tested under realistic life con-ditions. It is shown that some unexpected discrepancies between predictions and experimental findings exist, suggesting that a better empirical basis is required for predictions. The following ten suggestions may improve our potential to scale up from experimental scales to the real world. (1) Experiments should be timed to account for non-linearity in system responsiveness, asynchrony of responses and developmental differences. (2) By altering mineral nutrient supply, a wide range of CO₂ responses can be ‘produced’, thus requiring realistic soil conditions. (3) Distinctions should be made between ‘doubling CO₂ sup-ply’ and biologically effective degrees of CO₂ enrichment. (4) Because of the non-linearity of plant responses to CO₂, studies of at least three instead of two CO2 concentrations are necessary to describe future trends adequately. (5) Edge effects, in particular unscreened side light, may lead to allometric anomalies, strongly constraining up-scaling to stand-scale CO₂ responses. (6) Variables such as growth, yield, net primary production and C turnover are often confused with carbon pools, carbon sequestration or net ecosystem production. (7) Mono- and interspecific interactions between individuals may lead to completely unpredictable CO₂ responses. (8) Experiments with seedlings benefit from the absence of prehistory effects but are likely to be irrelevant for the responses of larger trees which, on the other hand, may be constrained by carry-over effects. Tree ring research indicates immediate sensitivity of large trees to environmental changes, supporting their usefulness in short-term CO₂-enrichment experiments. (9) In predicting temperature responses, acclimation deserves more attention. (10) The significance of developmental responses is largely under-represented in experimental research, although these responses may overrule many of the other effects of atmospheric change. Results of more realistic experiments which account for these problems will provide a better basis for modelling the future of the biosphere.     

Effects of an induced drought on soil carbon dioxide (CO2) efflux and soil CO2 production in an Eastern Amazonian rainforest, Brazil

In the next few decades, climate of the Amazon basin is expected to change, as a result of deforestation and rising temperatures, which may lead to feedback mechanisms in carbon (C) cycling that are presently unknown. Here, we report how a throughfall exclusion (TFE) experiment affected soil carbon dioxide (CO₂) production in a deeply weathered sandy Oxisol of Caxiuanã (Eastern Amazon). Over the course of 2 years, we measured soil CO₂ efflux and soil CO₂ concentrations, soil temperature and moisture in pits down to 3 m depth. Over a period of 2 years, TFE reduced on average soil CO₂ efflux from 4.3±0.1 μmol CO₂ m⁻² s⁻¹ (control) to 3.2±0.1 μmol CO₂ m⁻² s⁻¹ (TFE). The contribution of the subsoil (below 0.5 m depth) to the total soil CO₂ production was higher in the TFE plot (28%) compared with the control plot (17%), and it did not differ between years. We distinguished three phases of drying after the TFE was started. The first phase was characterized by a translocation of water uptake (and accompanying root activity) to deeper layers and not enough water stress to affect microbial activity and/or total root respiration. During the second phase a reduction in total soil CO₂ efflux in the TFE plot was related to a reduction of soil and litter decomposers activity. The third phase of drying, characterized by a continuing decrease in soil CO₂ production was dominated by a water stress‐induced decrease in total root respiration. Our results contrast to results of a drought experiment on clay Oxisols, which may be related to differences in soil water retention characteristics and depth of rooting zone. These results show that large differences exist in drought sensitivity among Amazonian forest ecosystems, which primarily seem to be affected by the combined effects of texture (affecting water holding capacity) and depth of rooting zone.