Bradford et al. (2016). Managing uncertainty in soil carbon feedbacks to climate change

NATURE CLIMATE CHANGE

| VOL 6 | AUGUST 2016 |

www.nature.com/natureclimatechange

751

undreds of studies have shown the effects of warming on

soil-carbon (C) dynamics

1,2

. Much of this empirical research

has been motivated by the possibility that climate warming

will stimulate biologically mediated decomposition of soil C to

(refs 3–7). Enhanced rates of soil-C decomposition may reduce

the capacity of the land to act as a CO

sink, so that a greater propor-

tion of anthropogenic CO

emissions remain in the atmosphere

8–10

The magnitude of this so-called carbon–climate feedback is there-

fore critical for estimating the allowable greenhouse gas emissions

that are compatible with climate targets

. The soil is the largest store

of C (~1,500–2,400 PgC) in the terrestrial biosphere, containing

more than double the C of the atmosphere

11–13

. Hence, loss of even a

small proportion of this store may result in higher atmospheric-CO

concentrations and consequently additional planetary warming

8,14

Despite the wealth of research into warming effects on soil-C

dynamics, there is no consensus on the magnitude of warming-

induced reductions in soil C stocks

1,14,15

. The low confidence in the

projected range of soil-C losses arises in part from an empirical

focus on the responses of soil C decomposition rates to warming,

rather than the direct measurement of changes in total soil C stocks.

Confidence in projected losses is further eroded because emerging

ideas

2,16

about how soil C is formed and stabilized are not commonly

represented in the soil biogeochemical models used for climate

change projections

17–19

. Instead, the assumptions in these models

about the mechanisms underlying soil C responses to warming are

largely similar

11,14,20,21

and often conflict with emerging understand-

ing

16,22,23

. These assumptions underlie Earth system model (ESM)

projections of soil C losses through climate warming

9,24

The ESMs are the most complex of the climate models, and

incorporate the global C cycle to simulate how the atmosphere and

biosphere interact to shape climate trajectories

. The ESM projec-

tions of warming-induced soil C losses range from minimal, to

one-third of the stock lost by 2100

8,9,25

. The ESM characteristics

generating this wide range in the projected magnitude of the feed-

back are well documented, involving uncertainties in the parameter

values used to control the rate at which soil-C decomposes and the

sensitivity of this rate to warming

26,27

. Reducing these parameter

Managing uncertainty in soil carbon feedbacks to

climate change

Mark A. Bradford

1,2

*, William R. Wieder

3,4

Gordon B. Bonan

, Noah Fierer

5,6

Peter A. Raymond

and Thomas W. Crowther

1,2

Planetary warming may be exacerbated if it accelerates loss of soil carbon to the atmosphere. This carbon-cycle–climate

feedback is included in climate projections. Yet, despite ancillary data supporting a positive feedback, there is limited evidence

for soil carbon loss under warming. The low confidence engendered in feedback projections is reduced further by the common

representation in models of an outdated knowledge of soil carbon turnover. ‘Model-knowledge integration’ — representing in

models an advanced understanding of soil carbon stabilization — is the first step to build confidence. This will inform experi-

ments that further increase confidence by resolving competing mechanisms that most influence projected soil-carbon stocks.

Improving feedback projections is an imperative for establishing greenhouse gas emission targets that limit climate change.

uncertainties will do little to build confidence in the magnitude of

the modelled feedback.

In this Review Article, we distinguish the meaning of ‘uncertainty’

from ‘confidence’. Many forms of uncertainty exist when modelling

climate change and associated biosphere feedbacks

, but they do not

equally contribute to the confidence one has that projected changes

will occur

. We focus on the four major uncertainties underpinning

low confidence in the projections of soil C stock responses to warm-

ing. Three of these major uncertainties are empirical and the fourth

is related to how soil C turnover is treated in ESMs. Specifically, the

major modelling uncertainty is associated with representing com-

mon and outdated ideas about soil C turnover in the soil sub-models

of the ESMs. We demonstrate the importance of instead represent-

ing different ideas in ESMs — that is, ‘structural uncertainties’ —

that capture emerging concepts of soil C stabilization. To facilitate

these efforts, we propose collaborative ways forward for empiricists

and modellers to efficiently and rapidly improve confidence in pro-

jected soil-C–climate feedbacks.

Empirical uncertainties

We being by discussing, in turn, the three primary areas of empiri-

cal uncertainty that generate low confidence in feedback projec-

tions: (1) the paucity of direct observations of warming effects on

soil C stocks; (2) the potential for organism responses to warming

to alter short-term biogeochemical responses; and (3) the dramati-

cally changing ideas about how soil C formation and stabilization

are regulated.

Evidence for carbon loss. Empirical research into soil C stock

responses to warming has primarily focused on decomposition

(Fig. 1). There is compelling evidence from observational studies

across climate gradients, and both laboratory and field warming

experiments, that decomposition rates respond positively to warm-

ing (but see ref. 15), at least in the short-term (<1 to ~10 years)

1,2,30

These increases in decomposition of soil C to CO

occur with plants

present or absent, suggesting that warming accelerates C loss from

soils primarily by stimulating the activities of microbes

. This

School of Forestry and Environmental Studies, Yale University, New Haven, Connecticut 06511, USA.

Department of Terrestrial Ecology, Netherlands

Institute of Ecology (NIOO-KNAW), 6700 AB Wageningen, the Netherlands.

Climate and Global Dynamics Laboratory, National Center for Atmospheric

Research, Boulder, Colorado 80307, USA.

Institute for Arctic and Alpine Research, University of Colorado, Boulder, Colorado, 80309, USA.

Department

of Ecology and Evolutionary Biology, University of Colorado, Boulder, Colorado, 80309, USA.

Cooperative Institute for Research in Environmental

Sciences, University of Colorado, Boulder, Colorado 80309, USA. *e-mail:

mark.bradford@yale.edu

REVIEW ARTICLE

PUBLISHED ONLINE: 27 JULY 2016 |

DOI: 10.1038/NCLIMATE3071

2 0 1 6

M a c m i l l a n

P u b l i s h e r s

L i m i t e d .

A l l

r i g h t s

r e s e r v e d .

Page 2

752

NATURE CLIMATE CHANGE

| VOL 6 | AUGUST 2016 |

www.nature.com/natureclimatechange

microbial mechanism underlies the C-cycle pathway in the ESMs

through which soil C is redistributed to the atmosphere as the cli-

mate warms.

The soil C stock is not, however, just determined by microbial

decomposition rates. Changes in the soil C stock are the net prod-

uct of outputs (decomposition) and also inputs (soil C formation,

Fig. 1). More rapid decomposition is therefore not synonymous with

reductions in total soil C stocks

if coupled with similar increases

in soil C formation. The idea that soil C stocks are the net outcome

of inputs and outputs seems obvious and is captured by the soil

sub-models in the ESMs

. Yet expectations for reductions in soil C

under warming are still primarily driven by empirical data of accel-

erated decomposition rates

, despite little evidence that decomposi-

tion responses can be used to infer responses in soil C stocks

1,32

. In

a synthesis of field data, for example, the mean effect size of warm-

ing on decomposition rates was statistically significant and strongly

positive

. However, the same meta-analysis showed that the mean

effect of warming on soil C stocks was indistinguishable from zero.

Collectively, the plethora of studies reporting positive decomposi-

tion responses to warming provide weak, indirect support for the

existence of a positive land-C–climate feedback

1,5,6

The lack of direct evidence for reductions in soil C stocks may, in

part, reflect a signal-to-noise issue. Soil C varies markedly at nano-

metre to metre scales in quantity, chemistry and the physical setting

where it is found

33,34

. These attributes can protect soil C from micro-

bial decomposition, meaning that a large proportion of the C in a

given soil will respond slowly (if at all) to warming

. Looking for a

small change in a large, spatially variable stock makes it difficult to

quantify the effects of warming on total soil C (Fig. 1). For exam-

ple, the statistical power to detect a change in total soil C stocks at

a site is typically far below that recommended, demanding higher

replication than generally used

. The use of techniques (such as

isotopic labelling and soil pool fractionations) to track and quantify

C turnover in soil pools that are differently vulnerable, does offer a

solution for detecting a signal from among the noise

36–39

. However,

the issues with using such techniques to infer change in stock sizes

echoes those for decomposition; environmental change can alter the

sizes of individual C pools or fluxes without altering the total stock

The difficulties involved in detecting changes in the total size

of soil C stocks probably encouraged the use of indirect measure-

ments, such as decomposition rates, to understand warming effects.

However, demonstrating definitively that soil C stocks will be

reduced under warming requires a large number of sites, long time

scales (>20 years) and ecosystem (versus soil only) experimental

warming. Such long-term network data will not be available in the

short term

, but even collation of soil C stocks in existing field stud-

ies would be a step forward. So far, we know of only five published

field-warming studies that measured soil C directly under experi-

mental warming plots for timescales greater than ten years, and they

did not consistently show reductions in soil C

2,42

. Furthermore, of

the 34 studies that have compared soil C dynamics in control versus

experimental warming plots

2,42

— over both the short- and longer-

term — only six measured stocks. The remainder measured soil C

concentrations but these do not account for potential changes in

soil bulk density, which could markedly affect stock sizes

43,44

. A key

challenge then, is determining how best to improve confidence in

projected soil-C–climate feedbacks in lieu of the fact that there is

limited direct data on the effects of ecosystem warming on total soil

C stocks.

Organisms modify direct warming effects. Knowing how to best

represent organismal responses to climate change in biogeochemi-

cal models is a significant challenge

. Initial effects of chronic

disturbance on an ecosystem are often transient because the

Respiration

High

1 m

Erosion losses

Atmospheric CO

(from microbial respiration)

Plant carbon inputs

Atmospheric CO

Lateral transport of

dissolved carbon

Abo

ve gr

ound

Belo

w gr

ound

Micr

obial and plant

Soil-carbon content

Low

Soil-warming experiment

Figure 1 |

Soil C stocks are the net result of outputs and inputs of plant C, but most warming research focuses only on outputs, making stock responses

highly uncertain.

Warming-induced outputs (red arrows) in the schematic are represented as CO

fluxes, reflecting the assumption in ESMs that the

land-C–climate feedback occurs through warming stimulating the activities of soil microorganisms that decompose soil C. Losses of soil C do occur
through other pathways (lateral transport and soil erosion, blue and brown arrows), but warming effects on these losses are not well characterized.
Instead, the majority of warming studies focus on decomposition of soil C to CO

. Despite strong warming effects on decomposition, there are very few

observed reductions in soil C stocks. This paucity of data reflects the fact that there has been far less research into how warming affects soil C formation
through plant inputs (downward green arrows) versus its effects on decomposition. It also reflects the signal-to-noise issues in detecting a change in soil
C stocks, given marked local variation (horizontally as well as with depth) in soil C stocks and the fact only a proportion of this C is likely to be sensitive to
warming-induced losses.

REVIEW ARTICLE

NATURE CLIMATE CHANGE

DOI: 10.1038/NCLIMATE3071

2 0 1 6

M a c m i l l a n

P u b l i s h e r s

L i m i t e d .

A l l

r i g h t s

r e s e r v e d .

Page 3

NATURE CLIMATE CHANGE

| VOL 6 | AUGUST 2016 |

www.nature.com/natureclimatechange

753

organisms, whose activities mediate biogeochemical processes such

as decomposition, first respond physiologically and second through

changes in abundance (Fig. 2)

. Two decades of experimental sum-

mer warming of arctic tundra, for example, gradually increased the

dominance of woody plants, altering plant community architecture.

The altered plant community mitigated direct summer warming of

the soils but caused indirect warming in the winter. These longer-

term consequences stimulated plant C inputs at depth, increasing

both the activity of the soil microbes and soil C storage, despite the

fact that initial warming was considered to promote soil C loss

Soil microbial communities and controls on their activities also

shift as temperatures change, altering their collective responses to

warming in the short- versus long-term

4,5,47–50

. Substantive debate

exists as to whether these shifts will influence soil C decomposition

rates

5,46,47,51

. Adding to this uncertainty, new efforts to incorporate

soil microbial processes in biogeochemical models reveal that the

manner in which they are represented means that simulated accli-

mation to warming can alternatively exacerbate or mitigate soil

C losses

4,52,53

Some organismal responses to warming are incorporated in ESM

formulations. For example, positive responses of plant production

to warming are expected in cold, high-latitude systems because

higher temperatures extend the growing season

. Most ESMs

then project an increase in land-C stocks at high latitudes because

increased plant-C inputs to soils more than offset the increases in

soil C decomposition rates

11,54

. Both model and empirical findings

therefore highlight that warming responses of organisms influenc-

ing either soil C inputs or outputs will probably alter direct effects

of warming on soil C stocks. The reality, however, is that we do not

know whether the collective effects of these organismal responses

under warming will amplify, dampen or scarcely influence direct

warming-induced changes in the global stock of soil C.

Changing ideas on soil carbon stabilization. Low temperature is

considered to be one of the dominant forces protecting soil C from

decomposition

. In permafrost soils (those that are ≤0 °C for greater

than two years) decomposition proceeds slowly because of limited

availability of liquid water

56,57

. As liquid water becomes available,

microbial decomposition of soil C initially proceeds slowly because

cool temperatures directly limit activity

. Warming then releases

temperature limitation on the catalytic activities of intracellular and

extracellular microbial enzymes, accelerating decomposition of soil

C to CO

. The land-C–climate feedback in ESMs is primarily based

on the assumption of this fundamental biochemical response

. But a

paradigmatic shift in our understanding of how soil C is stabilized

casts doubt on whether such cellular processes can be directly scaled

to biosphere-atmosphere interactions driving the C cycle.

Microorganisms have been considered the primary agents of soil

C decomposition for over a century. They are now also recognized,

somewhat paradoxically, as dominant agents of soil C formation

(Fig. 3)

16,23,61

. As much as 80% of the soil C in mineral soils that is pro-

tected from decomposition through physico-chemical mechanisms,

exists in the form of microbial necromass and products

, and the

proportion may be higher at depth

. Soil microbes therefore convert

large fractions of plant-C inputs to CO

and a smaller fraction into

stable soil C

. The process has been likened to a microbial funnel,

whereby microorganisms consume unprotected C and a portion that

passes through is converted into decomposition-resistant forms

36,50,64

This dual role for microbes raises the possibility that warming could

accelerate the decomposition and stabilization of soil C, shifting

stocks toward proportionally more protected forms of C that are less

sensitive to warming (Fig. 3)

The emerging paradigm of soil C formation emphasizes micro-

bial growth efficiencies and mineral-matrix interactions as dominant

forces stabilizing soil C

16,23,66,67

. Structural plant compounds such as

lignin, previously considered resistant to microbial decomposition

66,68

represent a poor-quality substrate for microbial growth. Physiological

inefficiencies involved in growing on poor substrates result in more

plant C being respired to CO

instead of being transformed to micro-

bial biomass

. Under this paradigm, most stable C in mineral soils is

produced via microbial uptake of primarily metabolic plant matter

inputs, which microbes rapidly decompose and convert to biomass

efficiently (Fig. 3)

23,69

. The dominant pathway for these inputs may be

via plant roots and their associated mycorrhizal fungi

70,71

. If this para-

digm is applicable across multiple ecosystems, it could help explain

why increased inputs of structural plant C to soils, from aboveground

litter sources, may not translate to higher soil C stocks

40,72

Once microbial products are formed, interactions with mineral-

soil surfaces, such as clays, are required to protect them from decom-

position

22,34

. Hence, even in wet tropical rainforests, large stores of soil

C can be found in aerobically respiring mineral soils where decom-

position should otherwise be rapid

. The most vulnerable pools of

soil C to warming are likely to be those in organic-rich soils, such

as wetlands and in permafrost, where the lack of oxygen and liquid

water, respectively, slow microbial decomposition. In such systems

much of the soil C has not passed through the microbial funnel, but

exists as relatively undecomposed plant material

. With warming,

C in these soils will increasingly pass through the funnel, resulting

in a major fraction being decomposed to CO

and a small fraction

becoming microbial products available for ‘restabilization’. Yet wet-

land and permafrost soils are poorly represented in ESM simulations

of land-C–climate feedbacks (but see ref. 73) and there is little data

available to estimate the likely magnitude of their response to climate

change

11,58,74

. Given that permafrost contains as much C as all non-

permafrost soils combined

, such limitations must be addressed

given the potential importance of their responses in dictating the

magnitude of land-C–climate feedbacks

Uncertainties in modelling

The major modelling uncertainty generating low confidence in the

magnitude of projected soil C losses arises because of the common

Time since warming began

0 y

20 y

Certainty in temperature response of soil-carbon stock

High

Low

Soil

microbes

Plants

Certainty in temperature response of plants and microbes

High

Low

Physiological

adjustment

Population

turnover

Community

turnover

Figure 2 |

Timescale of organismal responses to warming, with the

potential that initial increases in microbial activity are exacerbated
or mitigated through physiological, population and community-
level responses as the warming perturbation continues.

Modelled

land-C–climate feedbacks rely on warming-mediated increases in the
potential activities of microbial enzymes, which catalyse the decomposition
of soil C, being maintained in the longer term. Yet, physiological
acclimation and turnover in populations and communities — both above
ground and below ground (represented as plants and soil microbes,
respectively) — may modify the assumed translation of this initial cell-level
warming response directly to changes in soil C stocks. How the responses
at intermediate levels of biological organization modify this translation
becomes increasingly uncertain with time, given the large and diverse array
of interactions that can occur to restructure communities.

REVIEW ARTICLE

NATURE CLIMATE CHANGE

DOI: 10.1038/NCLIMATE3071

2 0 1 6

M a c m i l l a n

P u b l i s h e r s

L i m i t e d .

A l l

r i g h t s

r e s e r v e d .

Page 4

754

NATURE CLIMATE CHANGE

| VOL 6 | AUGUST 2016 |

www.nature.com/natureclimatechange

and outdated assumptions about soil C turnover represented in the

soil sub-models of the ESMs. We dicuss these assumptions below

and how new advances in soil biogeochemical modelling suggest a

productive way forward.

Model structure. The soil sub-models in ESMs represent soil C

responses to warming in a common manner. Soil-C decomposition

to CO

follows a single first-order response curve, similar to half-

life plots for radioactive decay, where the time taken for a constant

fraction of soil C to decompose decreases with warming

21,27,75,76

. The

mechanism then assumes that climate warming increases the short

and long-term potential for microbes to decompose soil C to CO

(refs 8–10), presupposing a positive C–climate feedback

This representation of soil C turnover in ESMs has remained

essentially unchanged for two decades

20,21

. Over the same time,

advances in climate change projection have been made by repre-

senting the different mechanisms thought to underlie responses of

the physical climate system to anthropogenic emissions

29,77,78

. For

the physical climate, the spread in an ensemble of model projec-

tions with the same scenario forcing is taken as a measure of model

uncertainty

28,29,74,78

. A similar principle has been applied to C-cycle

projections in ESMs

9,11,14,24

but the validity of doing so is question-

able. Specifically, the use of multi-model ensembles underlies the

Coupled Model Intercomparison Project (CMIP), which is a hall-

mark of the Intergovernmental Panel on Climate Change (IPCC)

assessment reports

. A key objective of the CMIP is to quantify the

influence of structural uncertainty, reflected in representing differ-

ent mechanisms among models, on projected climate change

28,29,77

Yet, because they represent common mechanisms for soil C turno-

ver, the broad spread among ESMs in the magnitude of projected

land-C–climate feedbacks

8–10,74

is not the consequence of struc-

tural uncertainty

26,54

Differences among soil sub-model projections instead result

largely from parameter uncertainty. Values for parameters such as

the ‘decay constant’ for soil C, contribute to a sixfold difference in

the simulated global stock of contemporary soil C

. Differences

in the simulated stocks carry forward and translate to substantial

among-model variation in the strength of C–climate feedback

projections

26,54,79

. Efforts to refine parameter estimates may reduce

among-model variation

but will not improve confidence in pro-

jected soil C stock responses to warming

26,81

. These improvements

will only come through representing the new ideas about the mech-

anisms regulating soil C turnover.

Representing structural uncertainty in soil processes in the

ESMs has the potential to drastically change projected terrestrial

C-cycle feedbacks. For example, beyond warming, the CO

fer-

tilization effect is a dominant biogeochemical feedback in ESMs

Increasingly, model structures represent the fact that the rate of soil

nitrogen (N) supply can strongly constrain plant growth responses

to elevated atmospheric CO

. In general, representing coupled

C-N biogeochemistry dampens the CO

fertilization effect on

plant productivity and reduces terrestrial C storage

83,84

. Given the

complexity of representing global C and N cycles, inclusion of ter-

restrial N dynamics may increase C-cycle uncertainty, but should

build confidence in model projections. Similarly, recent efforts to

represent structural uncertainty in soil C responses to warming in

soil biogeochemical models

19,85–87

suggest that divergent projections

of the C–climate feedback will be observed if these efforts are inte-

grated into CMIP exercises. We argue that such initiatives are nec-

essary to represent the true uncertainty associated with projecting

terrestrial biogeochemical responses to climate change.

Advances in soil biogeochemical models. Many of the ESMs simu-

late soil processes by using some of the most widely applied, soil bio-

geochemical models

. These ‘conventional’ soil models assume that

decomposition of soil C to CO

is a product of microbial activity, but

that microbes do not regulate the rate of soil C turnover

. Instead,

the control on turnover is exerted by factors such as the chemistry of

different soil C compounds

. These conventional models therefore

explicitly represent controls on microbial activity, but the microbes

themselves are considered ‘implicit’ to the dynamics. Recent major

advances in understanding soil C dynamics have come about in the

broader context of soil biogeochemical models (as opposed to the

narrow subset of these models used in the ESMs) by representing

explicitly how microbial physiology, biomass and enzyme kinetics

respond to warming

27,48,65

Marked reductions in the size of the soil C stock in response

to 30 years of simulated warming were observed with a conven-

tional model structure but not with a microbial-explicit structure,

despite the assumption that all of the processes were temperature

sensitive

. The difference in model structure was that, in the lat-

ter case, the microbial biomass controlled soil C decomposition

rates via production of degradative enzymes. As microbial biomass

and hence enzyme production declined over time, because warm-

ing was assumed to reduce microbial growth efficiencies

, decom-

position slowed and soil C stocks were maintained. Whereas the

conventional model projected that warming would reduce soil C

stocks, because the first-order decay mechanism assumes soil C

decomposition rates are independent of the size of the microbial

biomass

. Whether the microbial-explicit models are a more accu-

rate mechanistic representation of soil C dynamics is unknown.

However, they have been shown to improve the ability of conven-

tional soil C model structures to estimate observed spatial variation

and stock sizes of global soil C, as well as their responses to envi-

ronmental change

19,85–87

Another implicit assumption of most conventional soil mod-

els is that the rate of soil C formation (as opposed to decomposi-

tion) is regulated by microbial growth efficiencies, with higher

efficiencies leading to higher formation rates

. Warming-induced

reductions in growth efficiencies therefore exacerbate, rather than

mitigate as estimated by microbial-explicit models, losses of soil C

stocks

. Although microbial growth efficiencies are assumed to be

invariant in most conventional models (an assumption that is under

Soil microbes

Stable soil carbon

Plant

carbon

inputs

More warming

sensitive

soil carbon

Microbial product–mineral

surface interactions

Figure 3 |

The dual role of soil microbes as the agents of both soil C

decomposition and stabilization.

A new conceptualization of how soil

C is formed and stabilized emphasizes that plant-C inputs on which
microbes grow most efficiently result in larger protected stocks of soil C.
Warming-induced increases in decomposition rates may then cause more
unprotected (that is, more warming sensitive) soil C to be converted into
stable pools, ultimately mitigating the presumed land-C–climate feedback
because accelerated decomposition rates are balanced by elevated
formation rates. The grey arrow depicts this theoretical shunt of soil C from
more to less temperature-sensitive pools under warming via the microbial
funnel into microbial product–mineral complexes.

REVIEW ARTICLE

NATURE CLIMATE CHANGE

DOI: 10.1038/NCLIMATE3071

2 0 1 6

M a c m i l l a n

P u b l i s h e r s

L i m i t e d .

A l l

r i g h t s

r e s e r v e d .

Page 5

NATURE CLIMATE CHANGE

| VOL 6 | AUGUST 2016 |

www.nature.com/natureclimatechange

755

debate

4,90

), the example demonstrates that even slight structural

differences in how soil C turnover is represented can translate to a

broad spread in the projected magnitude of soil C losses. Until such

structural uncertainties are represented in ESM soil sub-models, we

cannot know whether the spread in the projected magnitude of the

C–climate feedback is reflective of our contemporary conceptual

understanding of soil C turnover. As such there is low confidence

in the current ESM projections of the C–climate feedback and the

resulting constraints on allowable greenhouse gas emissions

14,74

Addressing uncertainties to build confidence

We have neither the data nor the models required to reliably deter-

mine how soil C stocks will be affected by a warmer world. These

realities suggest a need for modellers and empiricists to collaborate

to increase confidence in the magnitude of projected C–climate

feedbacks. Such collaborations will succeed through an open

discussion — of the knowledge and data gaps in soil C research —

between the more geophysical-based Earth system modelling com-

munity and the more ecological-based empirical community

We propose four ways forward for modellers and empiricists

to focus efforts on identifying and addressing critical and tangible

assumptions that generate low confidence in projected soil C stock

responses. The overarching idea is to induce an exchange cycle of

model and empirical insights that rapidly advance mechanistic

understanding of how soil C is formed, stabilized and decomposed.

The expectation is that these advances in mechanistic understand-

ing will improve confidence in soil C stock responses to warming,

at timescales more attractive than those required to assemble direct

field observations of soil C stock responses to long-term, chronic

ecosystem warming. Importantly, the development of physical cli-

mate models shows that higher confidence in feedback projections

may be achieved through better representation of mechanisms, even

when this does not reduce the spread among model projections

We suggest that the aims for soil biogeochemistry should mirror

these developments, with the primary focus on representing and

improving our basic understanding of soil C cycling and a second-

ary focus on reducing the spread among models in the magnitude

of projected feedbacks (Fig. 4).

Represent structural uncertainty. The most important short-term

goal in Earth system modelling efforts focused on the land-C–

climate feedback should be to represent theoretical uncertainty in

soil C decomposition and formation processes through different

underlying model structures. The major advances in representing

such structural uncertainty in microbial-explicit soil biogeochemi-

cal models

48,52,63,65,92

should facilitate rapid adoption of competing

assumptions in soil sub-models embedded within ESMs. The fun-

damental approach should emulate standard practices in the atmos-

pheric sciences to sample model structural uncertainty through

multi-model ensembles

. The standardized protocols, however,

should more narrowly compare different mechanistic represen-

tations of soil C decomposition and formation within a common

modelling framework (for example, different soil biogeochemical

models within the same ESM). If these intracomparison

efforts are

computationally too expensive at the level of complexity of ESMs,

it is feasible to compare structurally distinct soil biogeochemical

models within the land models that are a component of ESMs

19,93

Adopting these systematic ‘intracomparisons’ will facilitate effective

model evaluation and improvement at regional to global scales.

Empiricists have two distinct roles to play in these model

intracomparison projects. The first is to work with modellers to

develop structural representations that best reflect the competing

conceptualizations of soil C turnover. We refer to this approach as

model-knowledge integration to distinguish it from the now ubiq-

uitous model-data synthesis efforts, which have failed to redress

the low confidence in soil-C–climate feedbacks. Model-knowledge

integration will probably involve synthesizing hundreds of

published mathematical and conceptual soil models into broad

classifications

, and summarizing the general processes (and ideally

their associated equations) that need to be represented in models

This methodology shows awareness of the human and computa-

tional costs related to representing structural uncertainty in climate

models, which demands that only the most plausible sets of mecha-

nisms are compared.

The second role for empiricists is to design and execute empirical

studies that distinguish which of the rapidly proliferating mecha-

nisms put forth to govern soil C formation and stabilization

16,22,23,60

are most consistent with field observations of soil C turnover. The

rapid growth in development and application of isotopic label-

ling and fractionation techniques that permit a “look inside” the

soil to resolve such dynamics as root-microbial-mineral interac-

tions

34,37,38,40

, highlight the potential for such approaches to refine the

sets of mechanisms that must be represented in models. Empiricists

can be guided in these efforts by the structural assumptions about

soil C turnover to which simulated soil C stock responses are most

responsive. An improved mechanistic understanding of soil C turn-

over can then go hand-in-hand with reductions in model structural

uncertainty and, consequently, increasing confidence in the magni-

tude of the projected feedback (Fig. 4).

Refine parameter estimates. Once structural uncertainty is rep-

resented and reduced, the aim to lessen the spread in the pro-

jected magnitude of the C–climate feedback by refining parameter

Year

Timeline and activities

Confidenc

Spr

ead among models

Empirical

Models

Feedback assessment

Measure soil C stocks

after long-term warming

Evaluate spatially-explicit

model structures

Refine parameter estimates

through empirical work

Represent improved

structural uncertainty and

parameter estimates

Refine structural uncertainty

through empirical work

Represent structural

uncertainty in models

Establish long-term, field

warming studies

Figure 4 |

Proposed activities to address low confidence in the projected

magnitude of carbon–climate feedbacks.

Shown is a timeline for major

initiatives that empiricists and modellers can act on to address the
low confidence and wide projected spread in soil C stock responses to
warming. Real-world confidence (inverted triangle) in the projected
magnitude of these responses increases when different assumptions
(that is, structural uncertainty) about the mechanisms governing soil C
turnover are represented in models, which then guides empirical research
to advance understanding of the mechanisms to which the models are
most sensitive. The trajectory of change in the spread among models
of the projected magnitude of the feedback is less assured and may
increase when structural uncertainty is represented. To establish allowable
emissions of greenhouse gases to meet specified climate targets, high
confidence in widely divergent projections is superior to low confidence
in a narrow range of projections, because policy can then be developed in
light of the knowledge that the best available science is considered.

REVIEW ARTICLE

NATURE CLIMATE CHANGE

DOI: 10.1038/NCLIMATE3071

2 0 1 6

M a c m i l l a n

P u b l i s h e r s

L i m i t e d .

A l l

r i g h t s

r e s e r v e d .

Page 6

756

NATURE CLIMATE CHANGE

| VOL 6 | AUGUST 2016 |

www.nature.com/natureclimatechange

estimates

26,79,81

will be of great value. The current ensemble of ESMs

vary markedly in the assumed values of soil parameters, such as the

sensitivity of decay constants to warming

. Similarly, variation in

even the parameter value of a single physiological process in the

microbial-explicit models, can mean the difference between large

versus no losses of soil C stocks under warming

. Notably, soil C

stocks are sensitive to a huge range of processes in these micro-

bial models, including community composition, enzyme activi-

ties, carbon use efficiency, microbial turnover and mineral surface

interactions

19,48,65,95,96

. Parameter values for these processes tend to

be poorly constrained by observations

39,56,80,90,94

. Hence, modellers

can provide critical guidance to empiricists by identifying processes

where modelled soil C stocks are strongly sensitive to the assumed

parameter value. Such guidance will allow empiricists to focus on

identifying the plausible range of values observed in nature and how

these values depend on environmental conditions

57,97

Consider spatially explicit processes. A guiding principle in devel-

oping climate models and ESMs has been to represent major pro-

cesses, within a model, in a mathematically uniform manner across

space. This principle has been effective for representing physical

atmospheric processes and also biological processes such as photo-

synthesis, where trade-offs in the balance between leaf respiratory

demands and photosynthetic rates constrain the set of mechanisms

and parameter values that can co-occur

. However, soil C turnover

is regulated by interacting physical and biological processes, with

the relative role of each process dependent on a complex suite of

environmental conditions that vary in space

65,73

. For example, the

recent focus on the role of microbial-mineral surface interactions

in soil C stabilization

, combined with a move away from ideas of

inherent chemical recalcitrance of plant inputs

16,66,68

, may fail to rep-

resent soil C turnover in organic soils where mineral surfaces are not

abundant

22,99

. Furthermore, there is evidence that mycorrhizal asso-

ciations are key arbiters of soil C stocks

37,100

, that plants can bypass

microorganisms to decompose soil C

, and that physico-chemical

sorption/ desorption processes, and not biology, may regulate soil

C turnover

. Clearly there are a growing number of different and

potentially important controls on soil C stocks that remain poorly

understood. As such, it is possible that no single structural represen-

tation of soil biogeochemistry will be effective at simulating soil C

turnover under warming across diverse landscapes.

The practical and computational costs of using different structural

representations in ESMs may be infeasible, but there needs to be an

appreciation that even a single model structure may need to repre-

sent dramatic shifts in the mechanisms underlying soil dynamics. As

such, the spatial and temporal scales at which models are parameter-

ized and applied must be carefully considered

. For example, effec-

tive simulation of soil C turnover may require different mechanisms

to operate as the location changes from one underlain by a mineral

to an organic soil. These different mechanisms may even be required

for the same location, where the turnover of C in organic horizons

and underlying mineral horizons are controlled by different fac-

tors

. Initial investigations of such possibilities will be best facili-

tated by controlled experiments that separately resolve C turnover

in soil C fractions and horizons, and then test whether these dynam-

ics are better represented by models that use a common or spatially

dependent mechanistic structure to simulate the total soil C stock.

Establish long-term warming experiments. Achieving real-world

confidence in model projections may ultimately demand direct

observations of soil C stock responses to climate change

. Even once

theoretical advances in understanding soil C turnover are widely

represented and refined in ESMs, the expectation that projections

should be compared to observations is likely to remain. Long-term

field ecosystem warming studies are therefore required that, given

the expectation that dominant controls on soil C turnover change

with space

, are organized into networks that facilitate adoption

of standard approaches for robust comparison among studies

Experimental field studies are not without limitations. Imposed

step-changes in temperature may not elicit the same responses as

the observed chronic rise in temperatures. In addition, the few exist-

ing long-term studies exemplify how even a detailed understanding

of the processes governing the formation and decomposition of dif-

ferent soil C fractions, may not permit accurate inferences about

how the soil C stock will respond in the longer term

. For example,

organismal responses can lead to surprises that modify direct effects

of warming, meaning that anticipated losses of soil C stocks may not

necessarily occur

30,42

. Such long-term studies are arguably too few

to benchmark model projections against with any real confidence.

To be of most value for determining allowable CO

emissions

to meet specified climate targets, field-warming experiments should

be initiated now and/or existing warming experiments extended to

facilitate observations of soil C stock responses. These studies must

be well replicated, warm the aboveground as well as soils, and be

of sufficiently long duration so that changes in soil C stocks can

be reliably assessed despite the ‘noise’ created by the slow turno-

ver of the total C stock and its pronounced fine-scale spatial vari-

ation in size (Fig. 1). Long durations will also permit organismal

responses to manifest at multiple temporal scales (Fig. 2), increas-

ing confidence that the observed responses might represent those

of natural systems. Similarly, if such long-term studies use isotopic

and fractionation approaches

36–39

to resolve the turnover of soil C of

differing sensitivities to warming, then inferences that shorter-term

responses can be used to estimate total C stock responses can be

validated. If long-term studies are broadly initiated, they may need

to measure soil C stocks on a mass basis, and not simply C concen-

trations and/or depth-dependent stocks given their inherent limita-

tions

43,44

. These studies will face the usual challenge of the limited

duration of grants to fund research and so will demand a longer-

term view of the value they offer in terms of improving confidence

in the effectiveness of greenhouse gas emission targets.

Conclusions

Major conceptual advances across the past 20 years in understand-

ing soil C turnover are not yet reflected in the way in which soil

biogeochemistry is represented in ESMs. Evaluating these advances

in ESMs will identify how this new knowledge might alter expected

responses of soil C stocks to climate change. In turn, insights from

the models will expedite gains in basic understanding by identi-

fying mechanisms that must be empirically researched before we

can accurately simulate soil C turnover. These activities may ini-

tially increase the spread in the projected magnitude of soil C stock

responses to warming, but should systematically improve confi-

dence in the projections by factoring conceptual uncertainties into

recommendations to manage human-induced changes in climate.

Received 1 September 2015; accepted 1 June 2016;

published online 27 July 2016

References

1. Conant, R. T.

et al.

Temperature and soil organic matter decomposition

rates — synthesis of current knowledge and a way forward.

Glob. Change Biol.

17, 3392–3404 (2011).

2. Lu, M.

et al.

Responses of ecosystem carbon cycle to experimental warming:

a meta-analysis.

Ecology

94, 726–738 (2013).

3. Dorrepaal, E.

et al.

Carbon respiration from subsurface peat accelerated by

climate warming in the subarctic.

Nature

460, 616–619 (2013).

4. Frey, S. D., Lee, J., Melillo, J. M. & Six, J. The temperature response of soil

microbial efficiency and its feedback to climate.

Nature Clim. Change

3, 395–398 (2013).

5. Karhu, K.

et al.

Temperature sensitivity of soil respiration rates enhanced by

microbial community response.

Nature

513, 81–84 (2014).

6. Melillo, J. M.

et al.

Soil warming, carbon–nitrogen interactions, and forest

carbon budgets.

Proc. Natl Acad. Sci. USA

108, 9508–9512 (2011).

REVIEW ARTICLE

NATURE CLIMATE CHANGE

DOI: 10.1038/NCLIMATE3071

2 0 1 6

M a c m i l l a n

P u b l i s h e r s

L i m i t e d .

A l l

r i g h t s

r e s e r v e d .

Page 7

NATURE CLIMATE CHANGE

| VOL 6 | AUGUST 2016 |

www.nature.com/natureclimatechange

757

7. Zhou, J.

et al.

Microbial mediation of carbon-cycle feedbacks to climate

warming.

Nature Clim. Change

2, 106–110 (2012).

8. Friedlingstein, P.

et al.

Uncertainties in CMIP5 climate projections due to

carbon cycle feedbacks.

J Clim.

27, 511–526 (2014).

9. Arora, V. K.

et al.

Carbon-concentration and carbon-climate feedbacks in

CMIP5 Earth System Models.

J Clim.

26, 5289–5314 (2013).

10. Jones, C.

et al.

Twenty-first-century compatible CO

emissions and airborne

fraction simulated by CMIP5 Earth system models under four representative

concentration pathways.

J Clim.

26, 4398–4413 (2013).

This study laid out the idea of ‘allowable emissions’, highlighting

the importance of terrestrial carbon cycle uncertainty in projecting

allowable greenhouse gas emissions that are compatible with specified

climate targets.

11. Ciais, P.

et al.

Climate Change 2013: The Physical Science Basis

(eds Stocker, T. F.

et al.

) 465–570 (IPCC, Cambridge Univ. Press, 2013).

12. Jobbágy, E. G. & Jackson, R. B. The vertical distribution of soil organic carbon

and its relation to climate and vegetation.

Ecol. Appl.

10, 423–436 (2000).

13. Tarnocai, C.

et al.

Soil organic carbon pools in the northern circumpolar

permafrost region.

Glob. Biogeochem. Cycles

23, GB2023 (2009).

14. Denman, K. L.

et al.

Climate Change 2007: The Physical Science Basis

(eds Solomon, S.

et al.

) 499–587 (IPCC, Cambridge Univ. Press, 2007).

15. Giardina, C. P., Litton, C. M., Crow, S. E. & Asner, G. P. Warming-related

increases in soil CO

efflux are explained by increased below-ground carbon

flux.

Nature Clim. Change

4, 822–827 (2014).

This study across an elevation gradient in a tropical forest showed that

the positive relationship between temperature and soil respiration

rates occurred not through expected direct warming effects on soil-C

decomposition but because of higher plant C inputs belowground.

16. Schmidt, M. W. I.

et al.

Persistence of soil organic matter as an ecosystem

property.

Nature

478, 49–56 (2011).

17. Davidson, D. A., Savage, K. E. & Finzi, A. C. A big-microsite framework for

soil carbon modeling.

Glob. Change Biol.

20, 3610–3620 (2014).

This opinion piece proposed a modular model structure to represent

the complexity of processes influencing soil C turnover, bringing

representations of soil C turnover in line with those of photosynthesis in

ecosystem and global models.

18. Wieder, W. R., Grandy, A. S., Kallenbach, C. M. & Bonan, G. B. Integrating

microbial physiology and physio-chemical principles in soils with the

MIcrobial–MIneral Carbon Stabilization (MIMICS) model.

Biogeosci.

11, 1147–1185 (2014).

19. 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).

20. Jenkinson, D. S., Adams, D. E. & Wild, A. Model estimates of CO

emissions

from soil in response to global warming.

Nature

351, 304–306 (1991).

21. Todd-Brown, K. E. O., Hopkins, F. M., Kivlin, S. N., Talbot, J. M.

& Allison, S. D. A framework for representing microbial decomposition in

coupled climate models.

Biogeochem.

109, 19–33 (2012).

22. Castellano, M. J., Mueller, K. E., Olk, D. C., Sawyer, J. E. & Six, J. Integrating

plant litter quality, soil organic matter stabilization, and the carbon saturation

concept.

Glob. Change Biol.

21, 3200–3209 (2015).

This opinion laid out a new conceptual model that integrates advances in

understanding of how microbial physiology controls soil C cycling, with

established physico-chemical principles that dictate whether physiological

responses influence soil C stocks.

23. Cotrufo, M. F., Wallenstein, M. D., Boot, C. M., Denef, K. & Paul, E. The

Microbial Efficiency-Matrix Stabilization (MEMS) framework integrates plant

litter decomposition with soil organic matter stabilization: do labile plant

inputs form stable soil organic matter?

Glob. Change Biol.

19, 988–995 (2013).

24. Friedlingstein, P.

et al.

Climate-carbon cycle feedback analysis: results from

the C

MIP model intercomparison.

J. Clim.

19, 3337–3353 (2006).

25. Cox, P. M.

et al.

Sensitivity of tropical carbon to climate change constrained

by carbon dioxide variability.

Nature

494, 341–344 (2013).

26. Exbrayat, J.-F., Pitman, A. J. & Abramowitz, G. Response of microbial

decomposition to spin-up explains CMIP5 soil carbon range until 2100.

Geosci. Model Dev.

7, 3481–3504 (2014).

27. Todd-Brown, K. E. O.

et al.

Causes of variation in soil carbon simulations

from CMIP5 Earth system models and comparison with observations.

Biogeosci.

10, 1717–1736 (2013).

28. Hawkins, E. & Sutton, R. The potential to narrow uncertainty in regional

climate predictions.

Bull. Am. Meteorol. Soc.

90, 1095–1107 (2009).

This study identified sources of uncertainty in physical climate

projections: revealing that at decadal timescales model uncertainty

is a dominant uncertainty source at regional and global scales,

and highlighting the large gains in certainty possible by refining

climate models.

29. Knutti, R. & Sedláček, J. Robustness and uncertainties in the new CMIP5

climate model projections.

Nature Clim. Change

3, 369–373 (2013).

This study showed that model spread (that is, uncertainty) in physical-

based climate projections changed little from the fourth to fifth

assessment report of the IPCC, yet the authors argued that confidence in

these projections should be substantially greater given major advances in

representing mechanistic understanding.

30. Melillo, J. M.

et al.

Soil warming and carbon-cycle feedbacks to the climate

system.

Science

298, 2173–2176 (2002).

31. Kirschbaum, M. U. F. The temperature dependence of organic-

matter decomposition — still a topic of debate.

Soil Biol. Biochem.

38, 2510–2518 (2006).

32. Torn, M. S., Vitousek, P. M. & Trumbore, S. E. The influence of nutrient

availability on soil organic matter turnover estimated by incubations and

radiocarbon modeling.

Ecosystems

8, 352–372 (2005).

33. Lehmann, J.

et al.

Spatial complexity of soil organic matter forms at

nanometre scales.

Nature Geosci

. 1, 238–242 (2008).

34. Vogel, C.

et al.

Submicron structures provide preferential spots for carbon

and nitrogen sequestration in soils.

Nature Commun.

5, 2947 (2014).

This empirical study showed that decomposition of plant C inputs into

more stable soil C fractions occurred preferentially via association with

mineral surfaces already clustered with organic matter, changing ideas

about how soil clay content relates to the potential of soils to sequester C.

35. Strickland, M. S., DeVore, J. L., Maerz, J. C. & Bradford, M. A. Grass

invasion of a hardwood forest is associated with declines in belowground

carbon pools.

Glob. Change Biol.

16, 1338–1350 (2010).

36. Bradford, M. A., Keiser, A. D., Davies, C. A., Mersmann, C. A.

& Strickland, M. S. Empirical evidence that soil carbon formation from

plant inputs is positively related to microbial growth.

Biogeochem.

113, 271–281 (2013).

37. Clemmensen, K. E.

et al.

Roots and associated fungi drive long-term carbon

sequestration in boreal forest.

Science

339, 1615–1618 (2013).

38. Keiluweit, M.

et al.

Mineral protection of soil carbon counteracted by root

exudates.

Nature Clim. Change

5, 588–595 (2015).

This empirical study showed that plant-root inputs could directly

liberate soil C from protective associations with minerals, bypassing

the presumed direct microbial role in decomposing this ‘stable’

soil C fraction.

39. Liang, C. & Balser, T. C. Warming and nitrogen deposition lessen microbial

residue contribution to soil carbon pool.

Nature Commun.

3, 1222 (2012).

40. Neff, J. C.

et al.

Variable effects of nitrogen additions on the stability and

turnover of soil carbon.

Nature

419, 915–917 (2002).

41. Torn, M. S.

et al.

A call for international soil experiment networks

for studying, predicting, and managing global change impacts.

SOIL

1, 575–582 (2015).

42. Sistla, S. A.

et al.

Long-term warming restructures Arctic tundra without

changing net soil carbon storage.

Nature

497, 615–618 (2013).

43. Gifford, R. M. & Roderick, M. L. Soil carbon stocks and bulk density: spatial

or cumulative mass coordinates as a basis of expression?

Glob. Change Biol.

9, 1507–1514 (2003).

This study showed how conventional soil sampling procedures might

fail to measure real changes in soil C stocks with time, and the authors

proposed that a mass-dependent method be broadly adopted to address

these issues.

44. Hopkins, D. W.

et al.

Soil organic carbon contents in long-term

experimental grassland plots in the UK (Palace Leas and Park Grass)

have not

changed consistently in recent decades.

Glob. Change Biol.

15, 1739–1754 (2009).

45. Schmitz, O. J.

et al.

Animating the carbon cycle.

Ecosystems

17, 344–359 (2014).

46. Reich, P. B. The carbon dioxide exchange.

Science

329, 774–775 (2010).

47. Bradford, M. A.

et al.

Thermal adaptation of soil microbial respiration to

elevated temperature.

Ecol. Lett.

11, 1316–1327 (2008).

48. Hagerty, S. B.

et al.

Accelerated microbial turnover but constant growth

efficiency with warming in soil.

Nature Clim. Change

4, 903–906 (2014).

49. Crowther, T. W. & Bradford, M. A. Thermal acclimation in widespread

heterotrophic soil microbes.

Ecol. Lett.

16, 469–477 (2013).

50. Crowther, T. W.

et al.

Biotic interactions mediate soil microbial feedbacks to

climate change.

Proc. Natl Acad. Sci. USA

112, 7033–7038 (2015).

51. Mahecha, M. D.

et al.

Global convergence in the temperature sensitivity of

respiration at ecosystem level.

Science

329, 838–840 (2010).

52. Allison, S. D., Wallenstein, M. D. & Bradford, M. A. Soil-carbon

response to warming dependent on microbial physiology.

Nature Geosci.

3, 336–340 (2010).

53. Allison, S. D. Modeling adaptation of carbon use efficiency in microbial

communities.

Front. Microbiol.

5, e571 (2014).

REVIEW ARTICLE

NATURE CLIMATE CHANGE

DOI: 10.1038/NCLIMATE3071

2 0 1 6

M a c m i l l a n

P u b l i s h e r s

L i m i t e d .

A l l

r i g h t s

r e s e r v e d .

Page 8

758

NATURE CLIMATE CHANGE

| VOL 6 | AUGUST 2016 |

www.nature.com/natureclimatechange

54. Todd-Brown, K. E. O.

et al.

Changes in soil organic carbon storage

predicted by Earth system models during the 21st century.

Biogeosci.

11, 2341–2356 (2014).

55. Davidson, E. A. & Janssens, I. A. Temperature sensitivity of soil carbon

decomposition and feedbacks to climate change.

Nature

440, 165–173 (2006).

56. Carvalhais, N.

et al.

Global covariation of carbon turnover times with climate

in terrestrial ecosystems.

Nature

514, 213–217 (2014).

57. Davidson, E. A., Samanta, S., Caramori, S. S. & Savage, K. The Dual Arrhenius

and Michaelis-Menten kinetics model for decomposition of soil organic matter

at hourly to seasonal time scales.

Glob. Change Biol.

18, 371–384 (2012).

58. Schuur, E. A. G.

et al.

Climate change and the permafrost carbon feedback.

Nature

520, 171–179 (2015).

59. Bradford, M. A. Thermal adaptation of decomposer communities in warming

soils.

Front. Microbiol.

4, e333 (2013).

60. Lehmann, J. & Kleber, M. The contentious nature of soil organic matter.

Nature

528, 60–68 (2015).

61. Miltner, A., Bombach, P., Schmidt-Brücken, B. & Kästner, M. SOM genesis:

microbial biomass as a significant source.

Biogeochem.

111, 41–55 (2012).

62. Liang, C. & Balser, T. C. Microbial production of recalcitrant organic

matter in global soils: implications for productivity and climate policy.

Nature Rev. Microbiol.

9, 75–77 (2010).

63. Ahrens, B., Braakhekke, M. C., Guggenberger, G., Schrumpf, M. & Reichstein,

M. Contribution of sorption, DOC transport and microbial interactions to

the

C age of a soil organic carbon profile: insights from a calibrated process

model.

Soil Biol. Biochem.

88, 390–402 (2015).

64. Grandy, A. S. & Neff, J. C. Molecular C dynamics downstream: the

biochemical decomposition sequence and its impact on soil organic matter

structure and function.

Sci. Total Environ.

404, 297–307 (2008).

65. Tang, J. & Riley, W. J. Weaker soil carbon–climate feedbacks resulting from

microbial and abiotic interactions.

Nature Clim. Change

5, 56–60 (2015).

This study showed that use of a dynamic, as opposed to the conventional

static, model structure to represent spatiotemporal dependencies in

temperature, microbial and mineral surface interactions, predicted weaker

but more variable soil-C–climate feedbacks.

66. Dungait, J. A. J., Hopkins, D. W., Gregory, A. S. & Whitmore, A. P. Soil

organic matter turnover is governed by accessibility not recalcitrance.

Glob. Change Biol.

18, 1781–1796 (2012).

67. Doetterl, S.

et al.

Soil carbon storage controlled by interactions between

geochemistry and climate.

Nature Geosci.

8, 780–783 (2015).

68. Marschner, B.

et al.

How relevant is recalcitrance for the stabilization of

organic matter in soils?

J. Plant. Nutr. Soil. Sci.

171, 91–110 (2008).

69. Crowther, T. W.

et al.

Environmental stress response limits microbial

necromass contributions to soil organic carbon.

Soil Biol. Biochem.

85, 153–161 (2015).

70. Högberg, P. & Read, D. J. Towards a more plant physiological perspective on

soil ecology.

Trends Ecol. Evol.

21, 548–554 (2006).

71. van Hees, P. A. W., Jones, D. L., Finlay, R., Godbold, D. L. & Lundström, U. S.

The carbon we do not see-the impact of low molecular weight compounds on

carbon dynamics and respiration in forest soils: a review.

Soil Biol. Biochem.

37, 1–13 (2005).

72. Pittelkow, C. M.

et al.

Productivity limits and potentials of the principles of

conservation agriculture.

Nature

517, 365–368 (2015).

73. Koven, C. D., Lawrence, D. M. & Riley, W. J. Permafrost carbon−climate

feedback is sensitive to deep soil carbon decomposability but not deep soil

nitrogen dynamics.

Proc. Natl Acad. Sci. USA

112, 3752–3757 (2015).

74. Collins, M.

et al.

Climate Change 2013: The Physical Science Basis

(eds Stocker, T. F.

et al.

) 1029–1136 (IPCC, Cambridge Univ. Press, 2013).

75. Sierra, C. A., Müller, M. & Trumbore, S. E. Models of soil organic matter

decomposition: the SoilR package, version 1.0.

Geosci. Model Dev. Discuss.

5, 1045–1060 (2012).

76. Xia, J., Luo, Y., Wang, Y.-P. & Hararuk, O. Traceable components of terrestrial

carbon storage capacity in biogeochemical models.

Glob. Change Biol.

19, 2104–2116 (2013).

77. Flato, G.

et al.

Climate Change 2013: The Physical Science Basi

(eds Stocker, T. F.

et al.

) 741–866 (IPCC, Cambridge Univ. Press, 2013).

78. Stocker, T. F.

et al

. in

Climate Change 2013: The Physical Science Basis

(eds Stocker, T. F. et al.) 33–115 (IPCC, Cambridge Univ. Press, 2013).

79. Wenzel, S., Cox, P. M., Eyring, V. & Friedlingstein, P. Emergent constraints

on climate-carbon cycle feedbacks in the CMIP5 Earth system models.

J. Geophys. Res.

119, 794–807 (2014).

80. Luo, Y.

et al.

Towards more realistic projections of soil carbon dynamics by

Earth System Models.

Glob. Biogeochem. Cycles

29, 40–56 (2015).

81. Luo, Y., Keenan, T. F. & Smith, M. Predictability of the terrestrial carbon cycle.

Glob. Change Biol.

21, 1737–1751 (2015).

82. Reich, P. B.

et al.

Nitrogen limitation constrains sustainability of ecosystem

response to CO

Nature

440, 922–925 (2006).

83. Wieder, W. R., Cleveland, C. C., Smith, W. K. & Todd-Brown, K. Future

productivity and carbon storage limited by terrestrial nutrient availability.

Nature Geosci.

8, 441–444 (2015).

84. Zhang, Q., Wang, Y. P., Matear, R. J., Pitman, A. J. & Dai, Y. J. Nitrogen and

phosphorous limitations significantly reduce future allowable CO

emissions.

Geophys. Res. Lett.

41, 632–637 (2014).

85. Hararuk, O., Smith, M. J. & Luo, Y. Microbial models with data-driven

parameters predict stronger soil carbon responses to climate change.

Glob. Change Biol.

21, 2439–2453 (2015).

86. Sulman, B. N., Phillips, R. P., Oishi, A. C., Shevliakova, E. & Pacala, S. W.

Microbe-driven turnover offsets mineral-mediated storage of soil carbon under

elevated CO

Nature Clim. Change

4, 1099–1102 (2014).

87. Wieder, W. R., Grandy, A. S., Kallenbach, C. M., Taylor, P. G. & Bonan, G. B.

Representing life in the Earth system with soil microbial functional traits in the

MIMICS model.

Geosci. Model Dev. Discuss.

8, 2011–2052 (2015).

88. Parton, W. J., Schimel, D. S., Cole, C. V. & Ojima, D. S. Analysis of

factors controlling soil organic matter levels in Great Plains grasslands.

Soil. Sci. Soc. Am. J.

51, 1173–1179 (1987).

89. Bonan, G. B., Hartman, M. D., Parton, W. J. & Wieder, W. R. Evaluating

litter decomposition in earth system models with long-term litterbag

experiments: an example using the Community Land Model version 4 (CLM4).

Glob. Change Biol.

19, 957–974 (2013).

90. Sinsabaugh, R. L., Manzoni, S., Moorhead, D. L. & Richter, A. Carbon use

efficiency of microbial communities: stoichiometry, methodology and

modelling.

Ecol. Lett.

16, 930–939 (2013).

91. Burd, A. B.

et al.

Terrestrial and marine perspectives on modeling organic

matter degradation pathways.

Glob. Change Biol.

22, 121–136 (2016).

92. Grant, R. F., Humphreys, E. R. & Lafleur, P. M. Ecosystem CO

and CH

exchange in a mixed tundra and a fen within a hydrologically diverse Arctic

landscape: 1. Modeling versus measurements.

J. Geophys. Res-Biogeosci.

120, 1366–1387 (2015).

93. Jones, C.

et al.

Global climate change and soil carbon stocks; predictions

from two contrasting models for the turnover of organic carbon in soil.

Glob. Change Biol.

11, 154–166 (2005).

94. Manzoni, S. & Porporato, A. Soil carbon and nitrogen mineralization: theory

and models across scales.

Soil Biol. Biochem.

41, 1355–1379 (2009).

95. German, D. P., Marcelo, K. R. B., Stone, M. M. & Allison, S. D. The Michaelis-

Menten kinetics of soil extracellular enzymes in response to temperature:

a cross-latitudinal study.

Glob. Change Biol.

18, 1468–1479 (2012).

96. Tucker, C. L., Bell, J., Pendall, E. & Ogle, K. Does declining carbon-use

efficiency explain thermal acclimation of soil respiration with warming?

Glob. Change Biol.

19, 252–263 (2013).

97. Suseela, V., Conant, R. T., Wallenstein, M. D. & Dukes, J. S. Effects of soil

moisture on the temperature sensitivity of heterotrophic respiration vary

seasonally in an old-field climate change experiment.

Glob. Change Biol.

18, 336–348 (2012).

98. Shipley, B., Lechowicz, M. J., Wright, I. & Reich, P. B. Fundamental

trade-offs generating the worldwide leaf economics spectrum.

Ecology

87, 535–541 (2006).

99. Manning, P.

et al.

Simple measures of climate, soil properties and plant

traits predict national-scale grassland soil carbon stocks.

J. Appl. Ecol.

52, 1188–1196 (2015).

100. Averill, C., Turner, B. L. & Finzi, A. C. Plant−decomposer competition for

nitrogen increases soil carbon storage in ecto- and ericoid-mycorrhizal

ecosystems.

Nature

505, 543–545 (2014).

Acknowledgements

This work was supported by grants from the US National Science Foundation

(DEB-1021098 and DEB-1457614). M.A.B. was supported by The Royal Netherlands

Academy of Arts and Sciences (Visiting Professors Programme); T.W.C. was supported

by the Yale Climate and Energy Institute and a Marie Skłodowska Curie Fellowship;

and W.R.W. was supported by grants from the US Department of Agriculture (NIFA

2015-67003-23485) and the US Department of Energy (TES DE-SC0014374). Thanks to

W.v.d.P. and F.C. for comments on an earlier draft of this script.

Author contributions

M.A.B. conceived the overall idea for this manuscript and together with T.W.C.

synthesized empirical information, and with W.R.W. modelling knowledge. M.A.B.,

W.R.W., G.B.B., N.F., P.A.W. and T.W.C. then co-developed the ideas and written material.

Additional information

Reprints and permissions information is available online at

www.nature.com/reprints

Correspondence and requests for materials should be addressed to M.A.B.

Competing financial interests

The authors declare no competing financial interests.

REVIEW ARTICLE

NATURE CLIMATE CHANGE

DOI: 10.1038/NCLIMATE3071

2 0 1 6

M a c m i l l a n

P u b l i s h e r s

L i m i t e d .

A l l

r i g h t s

r e s e r v e d .

Bradford et al. (2016). Managing uncertainty in soil carbon feedbacks to climate change

Page 1

Page 2

Page 3

Page 4

Page 5

Page 6

Page 7

Page 8

Document Outline