Calibrating the data logger…
Synthesizes results from over a thousand manipulative experiments to quantify how terrestrial carbon cycling responds to global change drivers, using log-response ratios and multi-model soil carbon simulations to characterize uncertainty and broad ecological patterns.
AI-generated synthesis. An AI-synthesized research primer that reads the publications, datasets, and documents tagged as this neighborhood's members and summarizes the work as a brief literature-review-style narrative.
Read it as a starting point — a synthesized map of what is in this neighborhood and how it connects, not as a peer-reviewed literature review. Cited papers are grounded; the synthesis is the model's reading of them.
Terrestrial ecosystems — the forests, grasslands, tundra, and meadows that cover Earth's land surface — exchange enormous amounts of carbon with the atmosphere every year. Plants pull carbon dioxide from the air through photosynthesis, while soils release it back through the respiration of roots and microbes. The balance between these flows determines whether a landscape acts as a carbon sink (storing more than it releases) or a carbon source. Understanding how global change drivers — warming temperatures, altered precipitation, rising CO2, and nitrogen deposition — shift that balance is one of the central problems in ecosystem science, and it matters acutely for places like the Gunnison Basin, where high-elevation soils store large amounts of carbon in cold, slowly decomposing organic layers.
Because no single field experiment can capture the full range of ecosystem responses across climates, soils, and species, researchers turn to meta-analysis: a quantitative way of synthesizing results from many independent studies to reveal general patterns. In a meta-analysis, each study contributes an effect size — most commonly a response ratio, which compares the mean of a treatment group (for example, warmed plots) to the mean of a control group. A response ratio greater than one means the treatment increased the measured variable; less than one means it decreased. Translating that ratio into a percentage change makes the result easy to communicate: a response ratio of 1.15 corresponds to a 15 percent increase. These tools let scientists ask whether, on average, warming increases soil respiration, or whether elevated CO2 reliably boosts plant growth, across hundreds of sites.
Comprehensive search of Web of Science databases for manipulative experiments on terrestrial carbon cycling responses to global change drivers, follow...
Comparative simulation study using five different soil organic carbon models (DAYCENT, CORPSE, MIMICS, MEND, RESOM) to project responses to warming an...
1. Database used in the article entitled "A meta-analysis of 1,119 manipulative experiments on terrestrial carbon-cycling responses to global change",...
Model output and meta-analysis data from model-experiment comparison that came out of INTERFACE workshop. Includes output from five soil carbon models...
Blood plasma samples were collected from advanced-stage NSCLC patients as part of a clinical study (PROPHETIC; NCT04056247). All clinical sites rece...
A closely related approach is to compare those experimental syntheses against process-based models — computer simulations that represent how microbes decompose litter, how minerals stabilize carbon, and how plants allocate growth. When models and experiments disagree, the disagreement points to gaps in our mechanistic understanding. For subalpine systems around Gothic, where snowpack, soil moisture, and growing-season length are all changing, these synthesis tools provide the only practical way to place local observations into a global context.
Early synthesis efforts in this field focused on extracting general signals from the growing library of manipulative experiments — studies in which researchers deliberately warm plots, add CO2, apply nitrogen, or alter precipitation and then measure the ecosystem response. A landmark contribution by Song and colleagues compiled 1,119 such experiments to examine how terrestrial carbon cycling responds to multiple global change drivers simultaneously, highlighting underrepresented regions including semi-arid ecosystems, tropical and subtropical forests, and Arctic tundra (Song et al., 2019).
Parallel work brought experiments and models into direct dialogue. Sulman and colleagues synthesized 147 long-term warming and litter-addition experiments alongside five different soil organic carbon models that represent microbial degradation and mineral stabilization in contrasting ways (Sulman et al., 2018). Together, these two studies established the modern template for the field: large experimental syntheses on one side, multi-model comparisons on the other, and an explicit effort to reconcile the two.
The clearest message from this body of work is that ecosystem responses to global change are far more variable, and far less predictable, than simple expectations suggest. The meta-analysis of more than a thousand manipulative experiments revealed that interactions among multiple global change drivers — warming combined with altered precipitation, or CO2 combined with nitrogen — remain poorly characterized in the very regions where carbon-climate feedbacks may be largest, including semi-arid lands, tropical forests, and Arctic tundra (Song et al., 2019). That gap matters because forecasts of future climate depend on how carbon storage in these underrepresented biomes will shift.
The model-experiment comparison reinforced this picture of uncertainty in striking ways. Every one of the five soil carbon models projected that warming would increase CO2 release and reduce soil carbon stocks, yet nearly one-third of the actual experiments observed decreases in CO2 flux under warming, and nearly half observed increases in soil carbon stocks (Sulman et al., 2018). In other words, the real world frequently does the opposite of what the models predict. Models also diverged substantially from one another in their responses to litter addition, with the differences traced largely to how each model handled unprotected soil carbon and microbial growth (Sulman et al., 2018).
Perhaps most sobering, the existing experimental measurements of CO2 flux and total soil carbon were not precise or consistent enough to rule out any of the competing models (Sulman et al., 2018). The variability among sites is so large that the data cannot yet tell us which mechanistic representation of soil processes is correct. Model behavior was, however, sensitive to soil texture and litter chemistry: soils with higher clay content showed muted responses to both warming and added litter, suggesting that mineral protection of carbon buffers some ecosystems more than others (Sulman et al., 2018).
Early synthesis work in the 2010s established that experiments and models disagree (Sulman et al., 2018), and the late-2010s pushed toward truly global compilations spanning thousands of experiments and dozens of drivers (Song et al., 2019). The frontier since 2020 has shifted toward filling the geographic and conceptual gaps these syntheses exposed: expanding measurements in semi-arid systems, tropical and subtropical forests, and Arctic and alpine tundra, and designing experiments that manipulate more than one driver at a time. Researchers are also calling for longer-term measurements of carbon inputs to soil and for experiments that separate mineral-protected from unprotected soil carbon, since these distinctions appear to drive most of the disagreement among models (Sulman et al., 2018).
Methodologically, the field is moving toward tighter integration of experiments with process-based models, using the experiments not just to test predictions but to discriminate among competing mechanistic hypotheses. New imaging and measurement technologies developed in adjacent areas of plant and ecosystem science are beginning to feed into this effort, expanding what can be quantified in the field and laboratory.
Several questions stand out for the next decade. Why do so many warming experiments show increases in soil carbon, contrary to what every major model predicts (Sulman et al., 2018)? How do multiple global change drivers interact in the regions least represented in current syntheses — semi-arid grasslands, tropical forests, and high-latitude and high-elevation tundra (Song et al., 2019)? What kinds of new measurements — particularly long-term records of carbon inputs and separate accounting of protected versus unprotected soil carbon pools — would finally allow experiments to discriminate among competing models? For the Gunnison Basin specifically, where subalpine meadows and alpine tundra hold deep carbon stocks under a rapidly changing snow regime, answering these questions locally will require coupling long-term monitoring at RMBL with the global synthesis frameworks these landmark papers established.
Song, J., Wan, S., Piao, S., et al. (2019). A meta-analysis of 1,119 manipulative experiments on terrestrial carbon-cycling responses to global change. Nature Ecology & Evolution. →
Sulman, B. N., Moore, J. A. M., Abramoff, R., et al. (2018). Multiple models and experiments underscore large uncertainty in soil carbon dynamics. Biogeochemistry. →