Skip to content
Ecological Intelligence

Ecological evidence.
Independent perspective.
Explore Wild Lands ↗

Ecological Intelligence / Subject

soil carbon

Peer-reviewed North America

Rewetting streams stored more soil carbon—and released more methane in one treatment

Evidence
Peer-reviewed
Geography
Northern New Mexico, United States
System
Riparian meadow / stream restoration / soil carbon

WHAT THE EVIDENCE SAYS

Researchers sampled riparian soils near reaches restored with plug-and-ponds or beaver dam analogs. After three years, particulate organic carbon at plug-and-pond sites increased from about 22.3 to 62.5 grams carbon per kilogram, while measured methane flux rose from about 0.14 to 1.73 micrograms carbon per square meter per second. Beaver dam analog sites did not show consistent carbon or moisture increases, perhaps in part because steeper slopes limited sustained wetting.

WHY IT MATTERS / OUR READ

Managers should budget for both intended soil processes and greenhouse-gas tradeoffs, and evaluate terrain before assuming methods with similar names produce similar hydrology.

CAVEATS / UNCERTAINTY

Preliminary calculations suggest the added methane did not outweigh soil-carbon benefits, but they are not a full life-cycle climate accounting. Observational sites differ in slope and restoration design, and soil-carbon fractions are not identical to durable whole-site sequestration.

SOURCE

Restoration Ecology / USDA Forest Service — Krichels et al. (2026)
View original evidence ↗

A northern New Mexico comparison found different carbon and methane responses from plug-and-ponds and beaver dam analogs.

Read the Wild Lands assessment →
Peer-reviewed Asia

Mountain forest maps predicted tree carbon better than soil carbon

Evidence
Peer-reviewed
Geography
Taohe Reserve, Gansu, China
System
Mountain forest carbon and remote sensing

WHAT THE EVIDENCE SAYS

Across ninety plots, soil made up 71.5 percent of measured ecosystem carbon and biomass 24.9 percent. Random forests predicted plot-scale biomass and total carbon better than soil or deadwood, and performed less reliably when extrapolated spatially.

WHY IT MATTERS / OUR READ

Canopy greenness is not a measurement of every carbon pool. Carbon plans need component sampling and explicit uncertainty for terrain and unobserved extremes.

CAVEATS / UNCERTAINTY

A specific mountain reserve and uneven plot coverage; the reserve-wide maps are relative spatial guides, not reliable per-pixel carbon inventories.

SOURCE

Frontiers in Plant Science — Zong et al. (2026)
View original evidence ↗

Gansu field plots expose uncertainty hidden beneath a mapped total.

Read the Wild Lands assessment →
Peer-reviewed North America

Prescribed-fire season altered soil-carbon responses in a California forest

Evidence
Peer-reviewed
Geography
Young mixed-conifer forest, California, United States
System
Prescribed fire; forest soils; carbon cycling

WHAT THE EVIDENCE SAYS

A peer-reviewed U.S. Forest Service-linked field study compared spring and fall prescribed fires in a young mixed-conifer forest. Surface soil-carbon fractions and microbial activity responded differently: fall burns showed sharper immediate reductions in particulate organic carbon, while spring treatment maintained or increased some measures over the following two years.

WHY IT MATTERS / OUR READ

Burn timing is an ecological treatment choice as well as an operational window. Fuel moisture, burn severity and management objectives must accompany subsequent soil monitoring before managers compare prescribed-fire results.

CAVEATS / UNCERTAINTY

This is one forest context and limited post-burn period; treatment intensity and fuel consumption can partly explain observed differences. Soil carbon alone cannot determine the best safe fire window.

SOURCE

Forest Ecology and Management · California prescribed-fire season study
View original evidence ↗

Fall and spring burns differed in near-surface carbon dynamics rather than delivering an interchangeable treatment.

Read the Wild Lands assessment →
Peer-reviewed Asia

Total soil carbon stayed similar while shrub expansion shifted it toward a less stable form

Evidence
Peer-reviewed
Geography
Tibetan Plateau, China
System
Alpine shrub-expansion system

WHAT THE EVIDENCE SAYS

The accepted research article found that shrub expansion slowed shoot-litter decomposition, increasing labile particulate organic carbon while decreasing more persistent mineral-associated organic carbon. Total soil organic carbon did not change substantially, but the composition shifted in a direction associated with lower long-term stability.

WHY IT MATTERS / OUR READ

A stable total-carbon number can conceal a meaningful change in carbon persistence. That matters for rangeland and climate interpretation because vegetation change can reorganize the quality and residence time of stored carbon before the headline stock changes.

CAVEATS / UNCERTAINTY

The result comes from a Tibetan Plateau shrub-expansion context and should not be generalized to all woody encroachment. Carbon fractions are indicators of persistence, not direct measurements of future centuries-long carbon retention.

SOURCE

Journal of Plant Ecology — accepted research article
View original evidence ↗

On the Tibetan Plateau, slower shrub-litter decomposition increased particulate organic carbon but reduced mineral-associated organic carbon—a compositional change that a total-SOC metric would miss.

Read the Wild Lands assessment →
North America

A restored marsh recovered microbial activity before it recovered its carbon stock

Evidence
Peer-reviewed comparative field study
Geography
Two freshwater marshes in eastern Ontario, Canada
System
Freshwater marsh / wetland

WHAT THE EVIDENCE SAYS

Nine sampling locations across one intact and one restored freshwater marsh spanned permanent, intermittent and non-inundated conditions. Permanent inundation was associated with lower microbial biomass-related ratios and enzyme activities. Despite the restored marsh having about half the soil carbon concentration of the intact marsh, microbial biomass and measured enzyme activities were similar between marshes.

WHY IT MATTERS / OUR READ

Process recovery and stock recovery can run on different timelines. A restoration can regain parts of microbial functioning while soil carbon remains far from the reference condition, so success metrics should distinguish activity, composition and accumulated stocks.

CAVEATS / UNCERTAINTY

Only two marshes and nine locations were studied, so restoration history and site differences are confounded. Similar microbial activity does not establish equivalent long-term carbon burial, and the authors explicitly note that higher activity could have positive or negative implications for continued carbon storage.

SOURCE

Freshwater Biology — primary study
View original evidence ↗

Two eastern-Ontario marshes had similar microbial biomass and enzyme activity even though the restored marsh held only about half the soil carbon of the intact marsh.

Read the Wild Lands assessment →
Peer-reviewed

On the Tibetan Plateau, more efficient microbes did not mean more soil carbon

Evidence
Peer-reviewed
Geography
Tibetan Plateau, China
System
Alpine soils across an aridity gradient

WHAT THE EVIDENCE SAYS

Large-scale soil sampling across a roughly 3,000-km Tibetan Plateau aridity gradient found microbial carbon-use efficiency increased with aridity even as soil organic carbon declined, producing a weak negative CUE–SOC relationship. The CUE increase was driven mainly by lower microbial respiration while growth changed little. Correlation, random-forest and structural-equation analyses linked SOC more strongly to physicochemical-protection indicators—including mineral-associated organic carbon, clay, calcium-bound carbon and microbial residue carbon—along with plant inputs and edaphic conditions.

WHY IT MATTERS / OUR READ

Soil carbon cannot be inferred from one appealing microbial efficiency metric. Across this gradient, whether carbon was physically and chemically stabilized mattered more than whether microbes retained a larger fraction of processed carbon. For land-carbon assessment and restoration, that argues for keeping mineral context, plant inputs and stabilization pools in the model rather than treating microbial efficiency as a universal shortcut.

CAVEATS / UNCERTAINTY

This is a spatial gradient study, not a manipulation that isolates causal pathways. Structural-equation and random-forest importance do not prove that any single stabilization mechanism caused the SOC pattern, and Tibetan Plateau soils and vegetation do not represent every dryland. Carbon-use-efficiency estimates are also method-dependent.

SOURCE

Global Change Biology — primary study
View original evidence ↗

Across a 3,000-km aridity gradient, microbial carbon-use efficiency increased as soil organic carbon declined; mineral association, clay, calcium-bound carbon and microbial residues tracked the carbon pattern more strongly.

Read the Wild Lands assessment →
Peer-reviewed

Soil microbes are putting carbon into minerals through an extracellular pathway we have mostly ignored

Evidence
Peer-reviewed
Geography
Cropland soils across China
System
Agricultural soils

WHAT THE EVIDENCE SAYS

Researchers developed an oxygen-isotope method to quantify microbial extracellular polymeric substances across a national-scale survey of Chinese cropland soils. Production rates were comparable to microbial cellular residues and were statistically associated with mineral-associated organic carbon. In carbon-labeling microcosms, extracellular-derived carbon entered the mineral-associated pool at rates comparable to or greater than carbon from cellular residues.

WHY IT MATTERS / OUR READ

The soil-carbon story is becoming less about a single pool of dead microbial biomass and more about multiple microbial pathways into persistent mineral associations. That matters for models and monitoring because a mechanism can be quantitatively important even when it is largely invisible to the proxy we usually measure.

CAVEATS / UNCERTAINTY

The national survey establishes associations and the incorporation experiment was conducted in microcosms. The study identifies a mechanism; it does not yet establish a field-management practice that reliably increases durable carbon storage, nor does it show that boosting extracellular polymers would produce net ecosystem carbon gains.

SOURCE

Nature Geoscience
View original evidence ↗

A China-wide cropland survey plus isotope-tracing microcosms found microbial extracellular polymers contributing to mineral-associated soil carbon at rates comparable to cellular residues.

Read the Wild Lands assessment →

Mangrove carbon increased along a seaward chronosequence—but total-stock differences were not statistically clear

Evidence
Peer-reviewed chronosequence field study
Geography
Southern China
System
Seaward-expanding mangrove forest

WHAT THE EVIDENCE SAYS

Across a seaward-expanding mangrove chronosequence, mean ecosystem carbon stocks increased from younger to older stands, while depth profiles and labile carbon fractions also shifted. The reported total carbon-stock differences among age classes, however, were not statistically significant.

WHY IT MATTERS / OUR READ

Carbon accumulation can change internally before a headline stock estimate separates cleanly among stand ages. Monitoring only total carbon may miss changes in where carbon is stored and how stable or labile those pools are.

CAVEATS / UNCERTAINTY

Chronosequences substitute space for time, and the work represents one mangrove setting. Sample size and local sediment dynamics limit generalization; higher mean carbon in older stands is not evidence that every planted or expanding mangrove will follow the same trajectory.

SOURCE

Fundamental Research
View original evidence ↗

A southern China chronosequence showed older mangrove stands with higher mean carbon stocks and changing carbon fractions, while total-stock differences among ages remained statistically uncertain.

Read the Wild Lands assessment →
Peer-reviewed

In cypress plantations, litter substrate explained more humification variation than the management label

Evidence
Peer-reviewed
Geography
Sichuan Basin, China
System
Subtropical Cupressus funebris plantations

WHAT THE EVIDENCE SAYS

A 1,277-day litterbag study compared four plantation stands under different management histories. Strip thinning and filling showed higher humus carbon and humification degree and less-negative humus-carbon change than artificial gaps in exploratory contrasts, while litter substrate explained the largest share of modeled humification variation.

WHY IT MATTERS / OUR READ

Belowground carbon pathways can respond differently from visible stand structure. For management evaluation, litter identity, microclimate and decomposition stage may matter as much as the categorical treatment name.

CAVEATS / UNCERTAINTY

Each management practice was represented by only one stand; plots were nested subsamples, not independent stand-level replicates. The authors explicitly treat treatment contrasts as exploratory, and several global management effects were non-significant, so the result cannot support a broad prescription for strip thinning.

SOURCE

Frontiers in Soil Science
View original evidence ↗

EvidencePeer-reviewedGeographySichuan Basin, ChinaSystemSubtropical Cupressus funebris plantations WHAT THE EVIDENCE SAYS A 1,277-day litterbag study compared four plantation stands under different management histories. Strip thinning and filling showed

Read the Wild Lands assessment →
Peer-reviewed North America

Canada’s aspen carbon model was strong above ground and weak below it

Evidence
Peer-reviewed
Geography
British Columbia, Alberta, Saskatchewan, Manitoba and Northwest Territories, Canada
System
Unmanaged boreal and temperate aspen-dominated forest stands

WHAT THE EVIDENCE SAYS

An Ecological Applications study compared Canada's CBM-CFS3 carbon model with observations from 64 unmanaged aspen-dominated stands monitored since 2000. The model explained roughly 88% of observed variation in aboveground biomass and performed moderately for standing dead wood, but it did not reproduce spatial variation in forest-floor or top mineral-soil carbon nearly as well. Residual patterns pointed to stand density, age, moisture, clay content, drought-related mortality and soil-temperature sensitivity as priorities for model refinement.

WHY IT MATTERS / OUR READ

For land managers, the practical point is pool-specific validation. A carbon model that performs well for live trees is not automatically validated for the whole ecosystem. If a stewardship, harvest, restoration or carbon-credit decision depends materially on forest-floor or soil carbon, local measurements and uncertainty bounds should remain part of the decision rather than inheriting confidence from the aboveground fit.

CAVEATS / UNCERTAINTY

The study covers 64 unmanaged aspen-dominated Canadian stands and cannot represent every forest type or management regime. Soil measurements were much less temporally rich than tree observations, topsoil depth and model architecture constrain comparison, and poor fit in one carbon pool does not mean the entire model is unusable.

SOURCE

Ecological Applications
View original evidence ↗

Across 64 long-monitored aspen stands, the national carbon model reproduced most aboveground biomass variation but performed much less reliably for forest-floor and mineral-soil pools.

Read the Wild Lands assessment →
Peer-reviewed Global

Planting forests can rebuild soil life and carbon—but the trajectory depends on what came before

Evidence
Peer-reviewed
Geography
Global; six continents
System
Afforested and reforested soils across former agricultural, degraded and other non-forest lands

WHAT THE EVIDENCE SAYS

A Nature Ecology & Evolution synthesis compared 1,158 paired afforestation or reforestation sites with reference conditions across six continents. Afforestation was associated with greater soil microbial biomass and topsoil carbon across stand development, while reforestation showed substantial early recovery of soil carbon and microbial biomass toward original-forest conditions during roughly the first three decades. The magnitude and persistence of gains varied with prior land condition, stand age and restoration pathway.

WHY IT MATTERS / OUR READ

The useful signal is that forest recovery is not just a canopy process: belowground biological condition can improve alongside carbon storage. But the starting state and restoration pathway matter, so the same planting effort can produce different soil trajectories. For project design, degraded-land recovery and long-term monitoring of soil function deserve explicit treatment rather than assuming tree cover is a sufficient proxy.

CAVEATS / UNCERTAINTY

This global synthesis combines heterogeneous chronosequences and paired-site studies rather than one standardized experiment. Older reforestation stands did not show a uniformly continuing increase in every soil metric, and the result should not be converted into a universal prescription to plant trees in naturally open ecosystems where afforestation can harm native biodiversity, fire regimes or water balance.

SOURCE

Nature Ecology & Evolution
View original evidence ↗

A global synthesis of 1,158 paired sites finds afforestation and reforestation generally increase soil microbial biomass and topsoil carbon, with especially strong gains on degraded land and different time courses after forest recovery.

Read the Wild Lands assessment →
Peer-reviewed

Mangrove carbon did not increase in one simple line with stand age

Evidence
Peer-reviewed
Geography
Coastal islands, Bangladesh
System
Afforested mangrove plantations

WHAT THE EVIDENCE SAYS

Across five stand ages from 12 to 44 years on six coastal islands, Sonneratia apetala plantations showed highest vegetation and total ecosystem carbon density in the oldest stands, while soil organic carbon followed a unimodal pattern and peaked in 26-year-old stands. Older stands also shifted more belowground carbon into deeper soil layers.

WHY IT MATTERS / OUR READ

Blue-carbon trajectories depend on which carbon pool and soil depth are measured. A single age–carbon curve can therefore hide redistribution among vegetation and soil pools, which matters for monitoring protocols and claims about restoration performance.

CAVEATS / UNCERTAINTY

This is a chronosequence rather than repeated long-term measurement of the same stands, and it focuses on one planted mangrove species. Carbon storage alone does not establish biodiversity recovery or broader restoration success.

SOURCE

Environmental Research · Original evidence
View original evidence ↗

Across six Bangladeshi islands, vegetation and whole-ecosystem carbon kept building while soil organic carbon peaked at an intermediate plantation age.

Read the Wild Lands assessment →

Keep exploring

Follow the systems. Find the connections.

WildlifeLandWaterOceansClimateCitiesOne HealthGovernanceMethods / Monitoring
Browse the archive →