Editing IPCC:AR6/SRCCL/Chapter-2 (section)

== 2.7 Plant and soil processes underlying land–climate interactions ==

<div id="article-2-7-plant-and-soil-processes-underlying-land-climate-interactions-block-1"></div>

Projecting future complex interactions between land and climate require ESMs. A growing number of studies suggested that many processes important for interactions between land and climate were missing in the CMIP5-class ESMs and that the DGVMs used tended to elevate CO <sub>2</sub> emission and removals ( ''high confidence'' ) (Busch and Sage 2017 <sup>[[#fn:r1927|1927]]</sup> ; Rogers et al. 2017 <sup>[[#fn:r1928|1928]]</sup> ; Anderegg et al. 2016 <sup>[[#fn:r1929|1929]]</sup> ; Tjoelker 2018 <sup>[[#fn:r1930|1930]]</sup> ; Sulman et al. 2014 <sup>[[#fn:r1931|1931]]</sup> ; Wieder et al. 2018 <sup>[[#fn:r1932|1932]]</sup> ; Davidson et al. 2006a <sup>[[#fn:r1933|1933]]</sup> ).

Ecosystem complexity stemming from the diversity of plants, animals and microbes, as well as their biological responses to gradual climate changes (e.g., adaptive migration) and disturbance events (e.g., extreme weather events, fire, pest outbreaks) (Section 2.2), are of potential importance. Of these processes, this section focuses on plant and soil processes as recent empirical work, including those explained in the following subsections, offers potential for improved model projections under warmer and CO <sub>2</sub> -rich futures.

The magnitude of future uptake and release of CO <sub>2</sub> and other GHGs by vegetation are among the greatest uncertainties (Ciais et al. 2013b <sup>[[#fn:r1934|1934]]</sup> ). One reason for this uncertainty stems from the lack of understanding of the mechanisms responsible for plant responses to increasing temperatures. The short- and long-term projections of gross photosynthesis responses to changes in temperature, CO <sub>2</sub> and nutrient availability vary greatly among the models (Busch and Sage 2017 <sup>[[#fn:r1935|1935]]</sup> ; Rogers et al. 2017 <sup>[[#fn:r1936|1936]]</sup> ). Net CO <sub>2</sub> exchange requires estimation of autotrophic respiration, which is another source of uncertainty in ESM projections (Malhi et al. 2011 <sup>[[#fn:r1937|1937]]</sup> ). The importance of plant acclimation of photosynthesis and respiration in understanding vegetation response to climate change is now widely recognised ( ''high confidence'' ) (Rogers et al., 2017 <sup>[[#fn:r1938|1938]]</sup> ; Tan et al., 2017 <sup>[[#fn:r2206|2206]]</sup> ; Tjoelker, 2018 <sup>[[#fn:r1939|1939]]</sup> ; Vanderwel et al., 2015 <sup>[[#fn:r1940|1940]]</sup> ) (Section 2.7.1). Acclimation is broadly defined as the biochemical, physiological, morphological or developmental adjustments within the lifetime of organisms that result in improved performance under the new condition. Acclimation often operates over a time span of days to weeks, and can mitigate the negative effects of climate change on organismal growth and ecosystem functions (Tjoelker 2018 <sup>[[#fn:r1941|1941]]</sup> ).

Soil carbon and microbial processes, which interact with plant responses to climate, represent another large source of uncertainty in model projections ( ''medium confidence'' ) (Sections 2.7.2, 2.7.3 and 2.7.4). Given the wide range of uncertainty associated with SOC size estimates, CMIP5 models use a wide range of starting SOC stocks from 510–3040 GtC (Todd-Brown et al. 2013 <sup>[[#fn:r1942|1942]]</sup> ). Soil microbial respiration is estimated to release 40–70 GtC annually from the soil to the atmosphere globally (Hawkes et al. 2017 <sup>[[#fn:r1943|1943]]</sup> ). Projections of changes in global SOC stocks during the 21st century by CMIP5 models also ranged widely, from a loss of 37 Gt to a gain of 146 Gt, with differences largely explained by initial SOC stocks, differing carbon input rates and different decomposition rates and temperature sensitivities (Todd-Brown et al. 2013 <sup>[[#fn:r1944|1944]]</sup> ). With respect to land–climate interactions, the key processes affecting SOC stocks are warming (which is expected to accelerate SOC losses through microbial respiration) and acceleration of plant growth (which increases inputs of carbon to soils). However, complex mechanisms underlying SOC responses to moisture regimes, carbon addition, and warming drive considerable uncertainty in projections of future changes in SOC stocks (Sulman et al. 2014 <sup>[[#fn:r1945|1945]]</sup> ; Singh et al. 2010 <sup>[[#fn:r1946|1946]]</sup> ; Wieder et al. 2018 <sup>[[#fn:r1947|1947]]</sup> ).

<span id="temperature-responses-of-plant-and-ecosystem-production"></span>
=== 2.7.1 Temperature responses of plant and ecosystem production ===

<div id="section-2-7-1-temperature-responses-of-plant-and-ecosystem-production-block-1"></div>

Climate-change responses of net ecosystem production cannot be modelled by simple instantaneous response functions because of thermal acclimation responses of plants and soil microbes, as well as delayed responses arising from interactions between plants and the soil ( ''high confidence'' ) (Slot et al. 2014 <sup>[[#fn:r1948|1948]]</sup> ; Rogers et al. 2017 <sup>[[#fn:r1949|1949]]</sup> ; Tan et al. 2017 <sup>[[#fn:r1950|1950]]</sup> ; Tjoelker 2018 <sup>[[#fn:r1951|1951]]</sup> ). Photosynthesis and respiration of component plant species exhibit different functional shapes among species (Slot et al. 2014 <sup>[[#fn:r1952|1952]]</sup> ), and carbon balance at the stand level is influenced by respiration of ecosystem biomass other than plants. Large uncertainty remains for thermal responses of bacteria and other soil organisms (Section 2.7.5). Bayesian statistical estimates of global photosynthesis and total ecosystem respirations suggest that they exhibit different responses to thermal anomalies during the last 35 years (Li et al. 2018b <sup>[[#fn:r1953|1953]]</sup> ).

Thermal responses of plant respiration, which consumes approximately one half of GPP, have not been appropriately incorporated in most ESMs (Davidson et al., 2006 <sup>[[#fn:r1954|1954]]</sup> ; Tjoelker, 2018 <sup>[[#fn:r1955|1955]]</sup> ). Assumptions associated with respiration have been a major source of uncertainty for ESMs at the time of AR5. In most existing models, a simple assumption that respiration doubles with each 10°C increase of temperature (i.e., Q10 = 2) is adopted, ignoring acclimation. Even a small error stemming from this assumption can strongly influence estimated net carbon balance at large spatial scales of ecosystems and biomes over the time period of multiple decades (Smith and Dukes 2013 <sup>[[#fn:r1956|1956]]</sup> ; Smith et al. 2016b <sup>[[#fn:r1957|1957]]</sup> ). In order to estimate more appropriate thermal response curves of respiration, a global database including data from 899 plant species has been compiled (Atkin et al. 2015 <sup>[[#fn:r1958|1958]]</sup> ), and respiration data from 231 plants species across seven biomes have been analysed (Heskel et al. 2016 <sup>[[#fn:r1959|1959]]</sup> ). These empirical data on thermal responses of respiration demonstrate a globally convergent pattern (Huntingford et al. 2017 <sup>[[#fn:r1960|1960]]</sup> ). According to a sensitivity analysis of a relatively small number of ESMs, a newly derived function of instantaneous responses of plant respiration to temperature (instead of a traditional exponential function of Q10 = 2) makes a significant difference in estimated autotrophic respiration especially in cold biomes (Heskel et al. 2016 <sup>[[#fn:r1961|1961]]</sup> ).

Acclimation results in reduced sensitivity of plant respiration with rising temperature, in essence, down regulation of warming-related increase in respiratory carbon emission ( ''high confidence'' ) (Atkin et al. 2015 <sup>[[#fn:r1962|1962]]</sup> ; Slot and Kitajima 2015 <sup>[[#fn:r1963|1963]]</sup> ; Tjoelker 2018 <sup>[[#fn:r1964|1964]]</sup> ). For example, experimental data from a tropical forest canopy show that temperature acclimation ameliorates the negative effects of rising temperature to leaf and plant carbon balance (Slot et al. 2014 <sup>[[#fn:r1965|1965]]</sup> ). Analysis of CO <sub>2</sub> flux data to quantify optimal temperature of net primary production of tropical forests also suggest acclimation potential for many tropical forests (Tan et al. 2017 <sup>[[#fn:r1966|1966]]</sup> ). Comparisons of models with and without thermal acclimation of respiration show that acclimation can halve the increase of plant respiration with projected temperature increase by the end of 21st century (Vanderwel et al. 2015 <sup>[[#fn:r1967|1967]]</sup> ).

It is typical that acclimation response to warming results in increases of the optimum temperature for photosynthesis and growth (Slot and Winter 2017 <sup>[[#fn:r1968|1968]]</sup> ; Yamori et al. 2014 <sup>[[#fn:r1969|1969]]</sup> ; Rogers et al. 2017 <sup>[[#fn:r1970|1970]]</sup> ). Although such shift is a result of a complex interactions of biochemical, respiratory and stomatal regulation (Lloyd and Farquhar 2008 <sup>[[#fn:r1971|1971]]</sup> ), it can be approximated by a simple algorithm to address acclimation (Kattge et al. 2007 <sup>[[#fn:r1972|1972]]</sup> ). Mercado et al. (2018) <sup>[[#fn:r1973|1973]]</sup> , using this approach, found that inclusion of biogeographical variation in photosynthetic temperature response was critically important for estimating future land surface carbon uptake. In the tropics, CO <sub>2</sub> fertilisation effect (Box 2.3) is suggested to be more important for observed increases in carbon sink strength than increased leaf area index or a longer growing season (Zhu et al. 2016 <sup>[[#fn:r1974|1974]]</sup> ). Acclimation responses of photosynthesis and growth to simultaneous changes of temperature and CO <sub>2</sub> , as well as stress responses above the optimal temperature for photosynthesis, remain a major knowledge gap in modelling responses of plant productivity under future climate change (Rogers et al. 2017 <sup>[[#fn:r1975|1975]]</sup> ).

<span id="water-transport-through-soil-plant-atmosphere-continuum-and-drought-mortality"></span>
=== 2.7.2 Water transport through soil-plant-atmosphere continuum and drought mortality ===

<div id="section-2-7-2-water-transport-through-soil-plant-atmosphere-continuum-and-drought-mortality-block-1"></div>

How climate change, especially changes of precipitation patterns, influence water transport through the soil-plant-atmosphere continuum, is a key element in projecting the future of water vapour flux from land and cooling via latent heat flux ( ''high confidence'' ) (Sellers et al. 1996 <sup>[[#fn:r1976|1976]]</sup> ; Bonan 2008 <sup>[[#fn:r1977|1977]]</sup> ; Brodribb 2009 <sup>[[#fn:r1978|1978]]</sup> ; Choat et al. 2012 <sup>[[#fn:r1979|1979]]</sup> ; Sperry and Love 2015 <sup>[[#fn:r1980|1980]]</sup> ; Novick et al. 2016 <sup>[[#fn:r1981|1981]]</sup> ; Sulman et al. 2016 <sup>[[#fn:r1982|1982]]</sup> ). Even without changes in leaf area per unit area of land, when plants close stomata in response to water shortage, dry atmosphere or soil moisture deficit, the stand-level fluxes of water (and associated latent heat flux) decrease (Seneviratne et al. 2018 <sup>[[#fn:r1983|1983]]</sup> ). Closing stomata enhances drought survival at the cost of reduced photosynthetic production, while not closing stomata avoids loss of photosynthetic production at the cost of increased drought mortality (Sperry and Love 2015 <sup>[[#fn:r1984|1984]]</sup> ). Hence, species-specific responses to drought, in terms of whether they close stomata or not, have short- and long-term consequences (Anderegg et al. 2018a <sup>[[#fn:r1985|1985]]</sup> ; Buotte et al. 2019 <sup>[[#fn:r1986|1986]]</sup> ). Increased drought-induced mortality of forest trees, often exacerbated by insect outbreak and fire (e.g., Breshears et al. (2005) <sup>[[#fn:r1987|1987]]</sup> , Kurz et al. (2008) <sup>[[#fn:r1988|1988]]</sup> , Allen et al. (2010) <sup>[[#fn:r1989|1989]]</sup> ) (Section 2.2.4), have long-term impact on hydrological interactions between land and atmosphere (Anderegg et al. 2018b <sup>[[#fn:r1990|1990]]</sup> ).

New models linking plant water transport with canopy gas exchange and energy fluxes are expected to improve projections of climate change impacts on forests and land-atmosphere interactions ( ''medium confidence'' ) (Bohrer et al., 2005 <sup>[[#fn:r1991|1991]]</sup> ; Anderegg et al., 2016 <sup>[[#fn:r1992|1992]]</sup> ; Sperry and Love, 2015 <sup>[[#fn:r1993|1993]]</sup> ; Wolf et al., 2016 <sup>[[#fn:r1994|1994]]</sup> ). Yet, there is much uncertainty in the ability of current vegetation and land surface models to adequately capture tree mortality and the response of forests to climate extremes like drought (Rogers et al. 2017 <sup>[[#fn:r1995|1995]]</sup> ; Hartmann et al. 2018 <sup>[[#fn:r1996|1996]]</sup> ). Most vegetation models use climate stress envelopes or vegetation carbon balance estimations to project climate-driven mortality and loss of forests (McDowell et al. 2011 <sup>[[#fn:r1997|1997]]</sup> ); these may not adequately project biome shifts and impacts of disturbance in future climates. For example, a suite of vegetation models was compared to a field drought experiment in the Amazon on mature rainforest trees and all models performed poorly in projecting the timing and magnitude of biomass loss due to drought (Powell et al. 2013 <sup>[[#fn:r1998|1998]]</sup> ). More recently, the loss of water transport due to embolism (disruption of xylem water continuity) (Sperry and Love 2015 <sup>[[#fn:r1999|1999]]</sup> ), rather than carbon starvation (Rowland et al. 2015 <sup>[[#fn:r2000|2000]]</sup> ), is receiving attention as a key physiological process relevant for drought-induced tree mortality (Hartmann et al. 2018 <sup>[[#fn:r2001|2001]]</sup> ). A key challenge to modelling efforts is to consider differences among plant species and vegetation types in their drought responses. One approach is to classify plant species to ‘functional types’ that exhibit similar responses to environmental variations (Anderegg et al. 2016 <sup>[[#fn:r2002|2002]]</sup> ). Certain traits of species, such as tree height, is shown to be predictive of growth decline and mortality in response to drought (Xu et al. 2016a <sup>[[#fn:r2003|2003]]</sup> ). Similarly, tree rooting depth is positively related to mortality, contrary to expectation, during prolonged droughts in tropical dry forest (Chitra-Tarak et al. 2017 <sup>[[#fn:r2004|2004]]</sup> ).

<span id="soil-microbial-effects-on-soil-nutrient-dynamics-and-plant-responses-to-elevated-co2"></span>
=== 2.7.3 Soil microbial effects on soil nutrient dynamics and plant responses to elevated CO2 ===

<div id="section-2-7-3-soil-microbial-effects-on-soil-nutrient-dynamics-and-plant-responses-to-elevated-co2-block-1"></div>

Soil microbial processes influencing nutrient and carbon dynamics represent a large source of uncertainty in projecting land–climate interactions. For example, ESMs incorporating nitrogen and phosphorus limitations (but without considering the effects of mycorrhizae and rhizosphere priming) indicate that the simulated future carbon-uptake on land is reduced significantly when both nitrogen and phosphorus are limited as compared to only carbon- stimulation, by 63% (of 197 Pg C) under RCP2.6 and by 67% (of 425 Pg C) under RCP8.5 (Zhang et al. 2013c <sup>[[#fn:r2005|2005]]</sup> ). Mineral nutrient limitation progressively reduces the CO <sub>2</sub> fertilisation effects on plant growth and productivity over time ( ''robust evidence, medium agreement'' ) (Norby et al. 2010 <sup>[[#fn:r2006|2006]]</sup> ; Sardans et al. 2012 <sup>[[#fn:r2007|2007]]</sup> ; Reich and Hobbie 2013 <sup>[[#fn:r2008|2008]]</sup> ; Feng et al. 2015 <sup>[[#fn:r2009|2009]]</sup> ; Terrer et al. 2017 <sup>[[#fn:r2010|2010]]</sup> ). The rates at which nutrient limitation develops differ among studies and sites. A recent meta- analysis shows that experimental CO <sub>2</sub> enrichment generally results in lower nitrogen and phosphorus concentrations in plant tissues (Du et al. 2019 <sup>[[#fn:r2011|2011]]</sup> ), and isotopic analysis also suggest a global trend of decreases in leaf nutrient concentration (Craine et al. 2018 <sup>[[#fn:r2012|2012]]</sup> ; Jonard et al. 2015 <sup>[[#fn:r2013|2013]]</sup> ). However, reduced responses to elevated CO <sub>2</sub> (eCO <sub>2</sub> ) may not be a simple function of nitrogen dilution per se, as they result from complex interactions of ecosystem factors that influence nitrogen acquisition by plants (Liang et al. 2016 <sup>[[#fn:r2014|2014]]</sup> ; Rutting 2017 <sup>[[#fn:r2015|2015]]</sup> ; Du et al. 2019 <sup>[[#fn:r2016|2016]]</sup> ).

Increasing numbers of case studies suggest that soil microbial processes, such as nitrogen mineralisation rates and symbiosis with plants, influence nutrient limitation on eCO <sub>2</sub> effects on plant growth ( ''medium confidence'' ) (Drake et al. 2011 <sup>[[#fn:r2017|2017]]</sup> ; Zak et al. 2011 <sup>[[#fn:r2018|2018]]</sup> ; Hungate et al. 2013 <sup>[[#fn:r2019|2019]]</sup> ; Talhelm et al. 2014 <sup>[[#fn:r2020|2020]]</sup> ; Du et al. 2019 <sup>[[#fn:r2021|2021]]</sup> ). Rhizosphere priming effects (i.e., release of organic matters by roots to stimulate microbial activities) and mycorrhizal associations are proposed to explain why some sites are becoming nitrogen limited after a few years and others are sustaining growth through accelerated nitrogen uptake (limited evidence, medium agreement) (Phillips et al. 2011 <sup>[[#fn:r2022|2022]]</sup> ; Terrer et al. 2017 <sup>[[#fn:r2023|2023]]</sup> ).

Model assessments that including rhizosphere priming effects and ectomycorrhizal symbiosis suggest that soil organic matter (SOM) cycling is accelerated through microbial symbiosis ( ''medium confidence'' ) (Elbert et al. 2012 <sup>[[#fn:r2024|2024]]</sup> ; Sulman et al. 2017 <sup>[[#fn:r2025|2025]]</sup> ; Orwin et al. 2011 <sup>[[#fn:r2026|2026]]</sup> ; Baskaran et al. 2017 <sup>[[#fn:r2027|2027]]</sup> ). Uncertainty exists in differences among ectomycorrhizal fungal species in their ability to decompose SOM (Pellitier and Zak 2018 <sup>[[#fn:r2028|2028]]</sup> ) and the capacity of ecosystems to sustain long-term growth with these positive symbiotic feedbacks is still under debate (Terrer et al. 2017 <sup>[[#fn:r2029|2029]]</sup> ). ESMs include only biological nitrogen cycles, even though a recent study suggests that bedrock weathering can be a significant source of nitrogen to plants (Houlton et al. 2018 <sup>[[#fn:r2030|2030]]</sup> ). In contrast, rock weathering is widely considered to be key for phosphorus availability, and tropical forests with highly weathered soils are considered to be limited by phosphorus availability rather than nitrogen availability (Reed et al. 2015 <sup>[[#fn:r2031|2031]]</sup> ). Yet evidence from phosphorus fertilisation experiments is lacking (Schulte-Uebbing and de Vries 2018 <sup>[[#fn:r2032|2032]]</sup> ) and phosphorus limitation of tropical tree growth may be strongly species-specific (Ellsworth et al. 2017 <sup>[[#fn:r2033|2033]]</sup> ; Turner et al. 2018a <sup>[[#fn:r2034|2034]]</sup> ). Limitation by availability of soil nutrients other than nitrogen and phosphorus has not been studied in the context of land–climate interactions, except potassium as a potentially limiting factor for terrestrial plant productivity in interaction with nitrogen, phosphorus and hydrology (Sardans and Peñuelas 2015 <sup>[[#fn:r2035|2035]]</sup> ; Zhao et al. 2017 <sup>[[#fn:r2036|2036]]</sup> ; Wright et al. 2018 <sup>[[#fn:r2037|2037]]</sup> ).

Anthropogenic alteration of global and regional nitrogen and phosphorus cycles, largely through use of chemical fertilisers and pollution, has major implications for future ecosystem attributes, including carbon storage, in natural and managed ecosystems ( ''high confidence'' ) (Peñuelas et al. 2013 <sup>[[#fn:r2038|2038]]</sup> , 2017 <sup>[[#fn:r2039|2039]]</sup> ; Wang et al. 2017c <sup>[[#fn:r2040|2040]]</sup> ; Schulte- Uebbing and de Vries 2018 <sup>[[#fn:r2041|2041]]</sup> ; Yuan et al. 2018 <sup>[[#fn:r2042|2042]]</sup> ). During 1997–2013, the contribution of nitrogen deposition to the global carbon sink has been estimated at 0.27 ± 0.13 GtC yr <sup>–1</sup> , and the contribution of phosphorus deposition as 0.054 ± 0.10 GtC yr <sup>–1</sup> ; these constitute about 9% and 2% of the total land carbon sink, respectively (Wang et al. 2017c <sup>[[#fn:r2043|2043]]</sup> ). Anthropogenic deposition of nitrogen enhances carbon sequestration by vegetation (Schulte-Uebbing and de Vries 2018 <sup>[[#fn:r2044|2044]]</sup> ), but this effect of nitrogen deposition on carbon sequestration may be offset by increased emission of GHGs such as N <sub>2</sub> O and CH <sub>4</sub> (Liu and Greaver 2009 <sup>[[#fn:r2045|2045]]</sup> ). Furthermore, nitrogen deposition may lead to imbalance of nitrogen vs phosphorus availability (Peñuelas et al. 2013 <sup>[[#fn:r2046|2046]]</sup> ), soil microbial activity and SOM decomposition (Janssens et al. 2010 <sup>[[#fn:r2047|2047]]</sup> ) and reduced ecosystem stability (Chen et al. 2016b <sup>[[#fn:r2048|2048]]</sup> ).

<span id="vertical-distribution-of-soil-organic-carbon"></span>
=== 2.7.4 Vertical distribution of soil organic carbon ===

<div id="section-2-7-4-vertical-distribution-of-soil-organic-carbon-block-1"></div>

It has long been recognised that dynamics of soil organic carbon (SOC) represent a large source of uncertainties on biogeochemical interactions of land with atmosphere and climate as detailed below. Since AR5, there have been new understandings on SOC size, as well as on the microbial processes that influence SOM dynamics under climate change and LULCC. Three existing databases (SoilGrids, the Harmonized World Soil Data Base and Northern Circumpolar Soil Database) substantially differ in the estimated size of global SOC stock down to 1 m depth, varying between 2500 Pg to 3400 Pg with differences among databases largely attributable to carbon stored in permafrost (Joosten 2015 <sup>[[#fn:r2049|2049]]</sup> ; Köchy et al. 2015 <sup>[[#fn:r2050|2050]]</sup> ; Tifafi et al. 2018 <sup>[[#fn:r2051|2051]]</sup> ). These values are four to eight times larger than the carbon stock associated with the terrestrial vegetation (Bond-Lamberty et al. 2018 <sup>[[#fn:r2052|2052]]</sup> ). New estimates since AR5 show that much larger areas in the Amazon and Congo basins are peatlands (Gumbricht et al. 2017 <sup>[[#fn:r2053|2053]]</sup> ; Dargie et al. 2019 <sup>[[#fn:r2054|2054]]</sup> ).

Deep soil layers can contain much more carbon than previously assumed ( ''limited evidence, medium agreement'' ) (e.g., González- Jaramillo et al. (2016) <sup>[[#fn:r2055|2055]]</sup> ). Based on radiocarbon measurements, deep SOC can be very old, with residence times up to several thousand years (Rumpel and Kögel-Knabner 2011 <sup>[[#fn:r2056|2056]]</sup> ) or even several tens of thousands of years (Okuno and Nakamura 2003 <sup>[[#fn:r2057|2057]]</sup> ). Dynamics associated with such deeply buried carbon remain poorly studied and ignored by the models, and are not addressed in most of the studies assessed in this subsection. Deep soil carbon is thought to be stabilised by mineral interactions, but recent experiments suggest that CO <sub>2</sub> release from deep soils can also be increased by warming, with a 4 ̊C warming enhancing annual soil respiration by 34–37% (Hicks Pries et al. 2017 <sup>[[#fn:r2058|2058]]</sup> ), or with the addition of fresh carbon (Fontaine et al. 2007 <sup>[[#fn:r2059|2059]]</sup> ). While erosion is not typically modelled as a carbon flux in ESMs, erosion and burial of carbon-containing sediments is likely a significant carbon transfer from land to ocean ( ''medium confidence'' ) (Berhe et al. 2007 <sup>[[#fn:r2060|2060]]</sup> ; Asefaw et al. 2008 <sup>[[#fn:r2061|2061]]</sup> ; Wang et al. 2017e <sup>[[#fn:r2062|2062]]</sup> ).

<span id="soil-carbon-responses-to-warming-and-changes-in-soil-moisture"></span>
=== 2.7.5 Soil carbon responses to warming and changes in soil moisture ===

<div id="section-2-7-5-soil-carbon-responses-to-warming-and-changes-in-soil-moisture-block-1"></div>

Annually, 119 GtC is estimated to be emitted from the terrestrial ecosystem to the atmosphere, of which about 50% is attributed to soil microbial respiration (Auffret et al. 2016 <sup>[[#fn:r2063|2063]]</sup> ; Shao et al. 2013 <sup>[[#fn:r2064|2064]]</sup> ). It is yet not possible to make mechanistic and quantitative projections about how multiple environmental factors influence soil microbial respiration (Davidson et al. 2006a <sup>[[#fn:r2065|2065]]</sup> ; Dungait et al. 2012 <sup>[[#fn:r2066|2066]]</sup> ). Soil warming experiments show significant variability in temperature and moisture responses across biomes and climates; Crowther et al. (2016) <sup>[[#fn:r2067|2067]]</sup> found that warming-induced SOC loss is greater in regions with high initial carbon stocks, while an analysis of an expanded version of the same dataset did not support this conclusion (Gestel et al. 2018 <sup>[[#fn:r2068|2068]]</sup> ). Studies of SOC responses to warming over time have also shown complex responses. In a multi-decadal warming experiment, Melillo et al. (2017) <sup>[[#fn:r2069|2069]]</sup> found that soil respiration response to warming went through multiple phases of increasing and decreasing strength, which were related to changes in microbial communities and available substrates over time. Conant et al. (2011) <sup>[[#fn:r2070|2070]]</sup> and Knorr et al. (2005) <sup>[[#fn:r2071|2071]]</sup> suggested that transient decomposition responses to warming could be explained by depletion of labile substrates, but that long-term SOC losses could be amplified by high temperature sensitivity of slowly decomposing SOC components. Overall, long-term SOC responses to warming remain uncertain (Davidson et al. 2006a <sup>[[#fn:r2072|2072]]</sup> ; Dungait et al. 2012 <sup>[[#fn:r2073|2073]]</sup> ; Nishina et al. 2014 <sup>[[#fn:r2074|2074]]</sup> ; Tian et al. 2015 <sup>[[#fn:r2075|2075]]</sup> ).

It is widely known that soil moisture plays an important role in SOM decomposition by influencing microbial processes (e.g., Monard et al. (2012) <sup>[[#fn:r2076|2076]]</sup> , Moyano et al. (2013) <sup>[[#fn:r2077|2077]]</sup> , Yan et al. (2018) <sup>[[#fn:r2078|2078]]</sup> ), as confirmed by a recent global meta-analysis ( ''high confidence'' ) (Hawkes et al. 2017 <sup>[[#fn:r2079|2079]]</sup> ). A likely mechanism is that increased soil moisture lowers carbon mineralisation rates under anaerobic conditions, resulting in enhanced carbon stocks, but experimental analyses have shown that this effect may last for only 3–4 weeks after which iron reduction can actually accelerate the loss of previously protected OC by facilitating microbial access (Huang and Hall 2017 <sup>[[#fn:r2080|2080]]</sup> ).

Experimental studies of responses of microbial respiration to warming have found variable results (Luo et al. 2001 <sup>[[#fn:r2081|2081]]</sup> ; Bradford et al. 2008 <sup>[[#fn:r2082|2082]]</sup> ; Zhou et al. 2011 <sup>[[#fn:r2083|2083]]</sup> ; Carey et al. 2016 <sup>[[#fn:r2084|2084]]</sup> ; Teramoto et al. 2016 <sup>[[#fn:r2085|2085]]</sup> ). No acclimation was observed in carbon-rich calcareous temperate forest soils (Schindlbacher et al. 2015 <sup>[[#fn:r2086|2086]]</sup> ) and arctic soils (Hartley et al. 2008 <sup>[[#fn:r2087|2087]]</sup> ), and a variety of ecosystems from the Arctic to the Amazon indicated that microbes appear to enhance the temperature sensitivity of soil respiration in Arctic and boreal soils, thereby releasing even more carbon than currently projected (Karhu et al. 2014 <sup>[[#fn:r2088|2088]]</sup> ). In tropical forests, phosphorus limitation of microbial processes is a key factor influencing soil respiration (Camenzind et al. 2018 <sup>[[#fn:r2089|2089]]</sup> ). Temperature responses of symbiotic mycorrhizae differ widely among host plant species, without a clear pattern that may allow generalisation across plant species and vegetation types (Fahey et al. 2016 <sup>[[#fn:r2090|2090]]</sup> ).

Some new insights have been obtained since AR5 from investigations of improved mechanistic understanding of factors that regulate temperature responses of soil microbial respiration. Carbon use efficiency and soil nitrogen dynamics have large influence on SOC responses to warming ( ''high confidence'' ) (Allison et al. 2010 <sup>[[#fn:r2091|2091]]</sup> ; Frey et al. 2013 <sup>[[#fn:r2092|2092]]</sup> ; Wieder, William R., Bonan, Gordon B., Allison 2013 <sup>[[#fn:r2093|2093]]</sup> ; García-Palacios et al. 2015 <sup>[[#fn:r2094|2094]]</sup> ). More complex community interactions including competitive and trophic interactions could drive unexpected responses to SOC cycling to changes in temperature, moisture and carbon inputs (Crowther et al. 2015 <sup>[[#fn:r2095|2095]]</sup> ; Buchkowski et al. 2017 <sup>[[#fn:r2096|2096]]</sup> ).

Competition for nitrogen among bacteria and fungi could also suppress decomposition (Averill et al. 2014 <sup>[[#fn:r2097|2097]]</sup> ). Overall, the roles of soil microbial community and trophic dynamics in global SOC cycling remain very uncertain.

<span id="soil-carbon-responses-to-changes-in-organic-matter-inputs-by-plants"></span>
=== 2.7.6 Soil carbon responses to changes in organic matter inputs by plants ===

<div id="section-2-7-6-soil-carbon-responses-to-changes-in-organic-matter-inputs-by-plants-block-1"></div>

While current ESM structures mean that increasing carbon inputs to soils drive corresponding increases in SOC stocks, long-term carbon addition experiments have found contradictory SOC responses. Some litter addition experiments have observed increased SOC accumulation (Lajtha et al. 2014b <sup>[[#fn:r2098|2098]]</sup> ; Liu et al. 2009 <sup>[[#fn:r2099|2099]]</sup> ), while others suggest insignificant SOC responses (Lajtha et al. 2014a <sup>[[#fn:r2100|2100]]</sup> ; van Groenigen et al. 2014 <sup>[[#fn:r2101|2101]]</sup> ). Microbial dynamics are believed to have an important role in driving complex responses to carbon additions. The addition of fresh organic material can accelerate microbial growth and SOM decomposition via priming effects (Kuzyakov et al. 2014 <sup>[[#fn:r2102|2102]]</sup> ; Cheng et al. 2017 <sup>[[#fn:r2103|2103]]</sup> ). SOM cycling is dominated by ‘hot spots’ including the rhizosphere as well as areas surrounding fresh detritus ( ''medium evidence, high agreement'' ) (Finzi et al. 2015 <sup>[[#fn:r2104|2104]]</sup> ; Kuzyakov and Blagodatskaya 2015 <sup>[[#fn:r2105|2105]]</sup> ). This complicates projections of SOC responses to increasing plant productivity as increasing carbon inputs could promote higher SOC storage, but these fresh carbon inputs could also deplete SOC stocks by promoting faster decomposition (Hopkins et al. 2014 <sup>[[#fn:r2106|2106]]</sup> ; Guenet et al. 2018 <sup>[[#fn:r2107|2107]]</sup> ; Sulman et al. 2014 <sup>[[#fn:r2108|2108]]</sup> ). A meta-analysis by van Groenigen et al. (2014) <sup>[[#fn:r2109|2109]]</sup> suggested that elevated CO <sub>2</sub> accelerated SOC turnover rates across several biomes. These effects could be especially important in high-latitude regions where soils have high organic matter content and plant productivity is increasing (Hartley et al. 2012 <sup>[[#fn:r2110|2110]]</sup> ), but have also been observed in the tropics (Sayer et al. 2011 <sup>[[#fn:r2111|2111]]</sup> ).

Along with biological decomposition, another source of uncertainty in projecting responses of SOC to climate change is stabilisation via interactions with mineral particles ( ''high confidence'' ) (Kögel- Knabner et al. 2008 <sup>[[#fn:r2112|2112]]</sup> ; Kleber et al. 2011 <sup>[[#fn:r2113|2113]]</sup> ; Marschner et al. 2008 <sup>[[#fn:r2114|2114]]</sup> ; Schmidt 2011 <sup>[[#fn:r2115|2115]]</sup> ). Historically, conceptual models of SOC cycling have centred on the role of chemical recalcitrance: the hypothesis that long-lived components of SOC are formed from organic compounds that are inherently resistant to decomposition. Under the emerging new paradigm, stable SOC is primarily formed by the bonding of microbially-processed organic material to mineral particles, which limits the accessibility of organic material to microbial decomposers (Lützow et al. 2006 <sup>[[#fn:r2116|2116]]</sup> ; Keiluweit et al. 2015 <sup>[[#fn:r2117|2117]]</sup> ; Kallenbach et al. 2016 <sup>[[#fn:r2118|2118]]</sup> ; Kleber et al. 2011 <sup>[[#fn:r2119|2119]]</sup> ; Hopkins et al. 2014 <sup>[[#fn:r2120|2120]]</sup> ). SOC in soil aggregates can be protected from microbial decomposition by being trapped in soil pores too small for microbes to access (Blanco-Canqui and Lal 2004 <sup>[[#fn:r2121|2121]]</sup> ; Six et al. 2004 <sup>[[#fn:r2122|2122]]</sup> ) or by oxygen limitation (Keiluweit et al. 2016 <sup>[[#fn:r2123|2123]]</sup> ). Some new models are integrating these mineral protection processes into SOC cycling projections (Wang et al. 2017a <sup>[[#fn:r2124|2124]]</sup> ; Sulman et al. 2014 <sup>[[#fn:r2125|2125]]</sup> ; Riley et al. 2014 <sup>[[#fn:r2126|2126]]</sup> ; Wieder et al. 2015 <sup>[[#fn:r2127|2127]]</sup> ), although the sensitivity of mineral-associated organic matter to changes in temperature, moisture, fire (Box 2.1) and carbon inputs is highly uncertain. Improved quantitative understanding of soil ecosystem processes will be critically important for projection of future land–climate feedback interactions.

<span id="section-2"></span>
<span id="footnotes"></span>