Editing IPCC:AR6/SRCCL/Chapter-4 (section)

== 4.3 Status and current trends of land degradation ==

<div id="article-4-3-status-and-current-trends-of-land-degradation-block-1"></div>

The scientific literature on land degradation often excludes forest degradation, yet here we attempt to assess both issues. Because of the different bodies of scientific literature, we assess land degradation and forest degradation under different sub-headings and, where possible, draw integrated conclusions.

<span id="land-degradation-1"></span>
=== 4.3.1 Land degradation ===

<div id="section-4-3-1-land-degradation-block-1"></div>

There are no reliable global maps of the extent and severity of land degradation (Gibbs and Salmon 2015 <sup>[[#fn:r421|421]]</sup> ; Prince et al. 2018 <sup>[[#fn:r422|422]]</sup> ; van der Esch et al. 2017 <sup>[[#fn:r423|423]]</sup> ), despite the fact that land degradation is a severe problem (Turner et al. 2016 <sup>[[#fn:r424|424]]</sup> ). The reasons are both conceptual – that is, how land degradation is defined, using what baseline (Herrick et al. 2019 <sup>[[#fn:r425|425]]</sup> ) or over what time period – and methodological – that is, how it can be measured (Prince et al. 2018 <sup>[[#fn:r426|426]]</sup> ). Although there is a strong consensus that land degradation is a reduction in productivity of the land or soil, there are diverging views regarding the spatial and temporal scales at which land degradation occurs (Warren 2002 <sup>[[#fn:r427|427]]</sup> ), and how this can be quantified and mapped. Proceeding from the definition in this report, there are also diverging views concerning ecological integrity and the value to humans. A comprehensive treatment of the conceptual discussion about land degradation is provided by the recent report on land degradation from the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES) (Montanarella et al. 2018 <sup>[[#fn:r428|428]]</sup> ).

A review of different attempts to map global land degradation, based on expert opinion, satellite observations, biophysical models and a database of abandoned agricultural lands, suggested that between &lt;10 Mkm2 to 60 Mkm2 (corresponding to 8–45% of the ice-free land area) have been degraded globally (Gibbs and Salmon, 2015 <sup>[[#fn:r429|429]]</sup> ) ( ''very low confidence'' ).

One often-used global assessment of land degradation uses trends in NDVI as a proxy for land degradation and improvement during the period 1983 to 2006 (Bai et al. 2008b <sup>[[#fn:r430|430]]</sup> ,c <sup>[[#fn:r431|431]]</sup> ) with an update to 2011 (Bai et al. 2015 <sup>[[#fn:r432|432]]</sup> ). These studies, based on very coarse resolution satellite data (NOAA AVHRR data with a resolution of 8 km), indicated that, between 22% and 24% of the global ice-free land area was subject to a downward trend, while about 16% showed an increasing trend. The study also suggested, contrary to earlier assessments (Middleton and Thomas 1997 <sup>[[#fn:r433|433]]</sup> ), that drylands were not among the most affected regions. Another study using a similar approach for the period 1981–2006 suggested that about 29% of the global land area is subject to ‘land degradation hotspots’, that is, areas with acute land degradation in need of particular attention. These hotspot areas were distributed over all agro-ecological regions and land cover types. Two different studies have tried to link land degradation, identified by NDVI as a proxy, and number of people affected: Le et al. (2016) <sup>[[#fn:r434|434]]</sup> estimated that at least 3.2 billion people were affected, while Barbier and Hochard (2016 <sup>[[#fn:r435|435]]</sup> , 2018 <sup>[[#fn:r436|436]]</sup> ) estimated that 1.33 billion people were affected, of which 95% were living in developing countries.

Yet another study, using a similar approach and type of remote-sensing data, compared NDVI trends with biomass trends calculated by a global vegetation model over the period 1982–2010 and found that 17–36% of the land areas showed a negative NDVI trend, while a positive or neutral trend was predicted in modelled vegetation (Schut et al. 2015 <sup>[[#fn:r437|437]]</sup> ). The World Atlas of Desertification (3rd edition) includes a global map of land productivity change over the period 1999 to 2013, which is one useful proxy for land degradation (Cherlet et al. 2018 <sup>[[#fn:r438|438]]</sup> ). Over that period, about 20% of the global ice-free land area shows signs of declining or unstable productivity, whereas about 20% shows increasing productivity. The same report also summarised the productivity trends by land categories and found that most forest land showed increasing trends in productivity, while rangelands had more declining trends than increasing trends (Figure 4.4). These productivity assessments, however, do not distinguish between trends due to climate change and trends due to other factors. A recent analysis of ‘greening’ of the world using MODIS time series of NDVI 2000–2017, shows a striking increase in the greening over China and India. In China the greening is seen over forested areas, 42%, and cropland areas, in which 32% is increasing (Section 4.9.3). In India, the greening is almost entirely associated with cropland (82%) (Chen et al. 2019 <sup>[[#fn:r439|439]]</sup> ).

All these studies of vegetation trends show that there are regionally differentiated trends of either decreasing or increasing vegetation. When comparing vegetation trends with trends in climatic variables, Schut et al. (2015 <sup>[[#fn:r440|440]]</sup> ) found very few areas (1–2%) where an increase in vegetation trend was independent of the climate drivers, and that study suggested that positive vegetation trends are primarily caused by climatic factors.

In an attempt to go beyond the mapping of global vegetation trends for assessing land degradation, Borelli et al. (2017) <sup>[[#fn:r441|441]]</sup> used a soil erosion model (RUSLE) and suggested that soil erosion is mainly caused in areas of cropland expansion, particularly in Sub-Saharan Africa, South America and Southeast Asia. The method is controversial for conceptual reasons (i.e., the ability of the model to capture the most important erosion processes) and data limitations (i.e., the availability of relevant data at regional to global scales), and its validity for assessing erosion over large areas has been questioned by several studies (Baveye 2017 <sup>[[#fn:r442|442]]</sup> ; Evans and Boardman 2016a <sup>[[#fn:r443|443]]</sup> ,b <sup>[[#fn:r444|444]]</sup> ; Labrière et al. 2015 <sup>[[#fn:r445|445]]</sup> ).

An alternative to using remote sensing for assessing the state of land degradation is to compile field-based data from around the globe (Turner et al. 2016 <sup>[[#fn:r446|446]]</sup> ). In addition to the problems of definitions and baselines, this approach is also hampered by the lack of standardised methods used in the field. An assessment of the global severity of soil erosion in agriculture, based on 1673 measurements around the world (compiled from 201 peer-reviewed articles), indicated that the global net median rate of soil formation (i.e., formation minus erosion) is about 0.004 mm yr <sup>–1</sup> (about 0.05 t ha <sup>–1</sup> yr <sup>–1</sup> ) compared with the median net rate of soil loss in agricultural fields, 1.52 mm yr <sup>–1</sup> (about 18 t ha <sup>–1</sup> yr <sup>–1</sup> ) in tilled fields and 0.065 mm yr <sup>–1</sup> (about 0.8 t ha–1 yr <sup>–1</sup> ) in no-till fields (Montgomery 2007a <sup>[[#fn:r447|447]]</sup> ). This means that the rate of soil erosion from agricultural fields is between 380 (conventional tilling) and 16 times (no-till) the natural rate of soil formation ( ''medium agreement, limited evidence'' ). These approximate figures are supported by another large meta-study including over 4000 sites around the world (see Figure 4.4) where the average soil loss from agricultural plots was about 21 t ha <sup>–1</sup> yr <sup>–1</sup> (García-Ruiz et al. 2015 <sup>[[#fn:r448|448]]</sup> ). Climate change, mainly through the intensification of rainfall, will further increase these rates unless land management is improved ( ''high agreement, medium evidence'' ).

<div id="section-4-3-1-land-degradation-block-2"></div>

<span id="figure-4.4"></span>
<!-- START IMG -->
<!-- IMG TITLE -->
'''Figure 4.4'''

<span id="proportional-global-land-productivity-trends-by-land-coverland-use-class.-cropland-includes-arable-land-permanent-crops-and-mixed-classes-with-over-50-crops-grassland-includes-natural-grassland-and-managed-pasture-land-rangelands-include-shrubland-herbaceous-and-sparsely-vegetated-areas-forest-land-includes-all-forest-categories-and-mixed-classes-with-tree-cover-greater-than-40.-data-source-copernicus"></span>
<!-- IMG CAPTION -->
'''Proportional global land productivity trends by land-cover/land-use class. (Cropland includes arable land, permanent crops and mixed classes with over 50% crops; grassland includes natural grassland and managed pasture land; rangelands include shrubland, herbaceous and sparsely vegetated areas; forest land includes all forest categories and mixed classes with tree cover greater than 40%.) Data source: Copernicus […]'''
<!-- IMG FILE -->
[[File:151937ca1e81119c0abacdbba868a370 Figure-4.4-1024x576.jpg]]

Proportional global land productivity trends by land-cover/land-use class. (Cropland includes arable land, permanent crops and mixed classes with over 50% crops; grassland includes natural grassland and managed pasture land; rangelands include shrubland, herbaceous and sparsely vegetated areas; forest land includes all forest categories and mixed classes with tree cover greater than 40%.) Data source: Copernicus Global Land SPOT VGT, 1999–2013, adapted from (Cherlet et al. 2018 <sup>[[#fn:r1647|1647]]</sup> ).

<!-- END IMG -->
<div id="section-4-3-1-land-degradation-block-3"></div>

Soils contain about 1500 Gt of organic carbon (median across 28 different estimates presented by Scharlemann et al. (2014)), which is about 1.8 times more carbon than in the atmosphere (Ciais et al. 2013 <sup>[[#fn:r449|449]]</sup> ) and 2.3–3.3 times more than what is held in the terrestrial vegetation of the world (Ciais et al. 2013 <sup>[[#fn:r450|450]]</sup> ). Hence, land degradation, including land conversion leading to soil carbon losses, has the potential to impact on the atmospheric concentration of CO <sub>2</sub> substantially. When natural ecosystems are cultivated they lose soil carbon that accumulated over long time periods.The loss rate depends on the type of natural vegetation and how the soil is managed. Estimates of the magnitude of loss vary but figures between 20% and 59% have been reported in several meta studies (Poeplau and Don 2015 <sup>[[#fn:r451|451]]</sup> ; Wei et al. 2015 <sup>[[#fn:r452|452]]</sup> ; Li et al. 2012 <sup>[[#fn:r453|453]]</sup> ; Murty et al. 2002 <sup>[[#fn:r454|454]]</sup> ; Guo and Gifford 2002 <sup>[[#fn:r455|455]]</sup> ). The amount of soil carbon lost explicitly due to land degradation after conversion is hard to assess due to large variation in local conditions and management, see also Chapter 2.

From a climate change perspective, land degradation plays an important role in the dynamics of nitrous oxide (N <sub>2</sub> O) and methane (CH <sub>4</sub> ). N <sub>2</sub> O is produced by microbial activity in the soil and the dynamics are related to both management practices and weather conditions, while CH <sub>4</sub> dynamics are primarily determined by the amount of soil carbon and to what extent the soil is subject to waterlogging (Palm et al. 2014 <sup>[[#fn:r456|456]]</sup> ), see also Chapter 2.

Several attempts have been made to map the human footprint on the planet (Čuček et al. 2012 <sup>[[#fn:r457|457]]</sup> ; Venter et al. 2016 <sup>[[#fn:r458|458]]</sup> ) but, in some cases, they confuse human impact on the planet with degradation. From our definition it is clear that human impact (or pressure) is not synonymous with degradation, but information on the human footprint provides a useful mapping of potential non-climatic drivers of degradation.

In summary, there are no uncontested maps of the location, extent and severity of land degradation. Proxy estimates based on remote sensing of vegetation dynamics provide one important information source, but attribution of the observed changes in productivity to climate change, human activities, or other drivers is hard. Nevertheless, the different attempts to map the extent of global land degradation using remotely sensed proxies show some convergence and suggest that about a quarter of the ice-free land area is subject to some form of land degradation ( ''limited evidence, medium agreement'' ) affecting about 3.2 billion people ( ''low confidence'' ). Attempts to estimate the severity of land degradation through soil erosion estimates suggest that soil erosion is a serious form of land degradation in croplands closely associated with unsustainable land management in combination with climatic parameters, some of which are subject to climate change ( ''limited evidence, high agreement'' ). Climate change is one among several causal factors in the status and current trends of land degradation ( ''limited evidence, high agreement'' ).

<span id="forest-degradation"></span>
=== 4.3.2 Forest degradation ===

<div id="section-4-3-2-forest-degradation-block-1"></div>

Quantifying degradation in forests has also proven difficult. Remote sensing based inventory methods can measure reductions in canopy cover or carbon stocks more easiliy than reductions in biological productivity, losses of ecological integrity or value to humans. However, the causes of reductions in canopy cover or carbon stocks can be many (Curtis et al. 2018 <sup>[[#fn:r459|459]]</sup> ), including natural disturbances (e.g., fires, insects and other forest pests), direct human activities (e.g., harvest, forest management) and indirect human impacts (such as climate change) and these may not reduce long-term biological productivity. In many boreal, some temperate and other forest types natural disturbances are common, and consequently these disturbance-adapted forest types are comprised of a mosaic of stands of different ages and stages of stand recovery following natural disturbances. In those managed forests where natural disturbances are uncommon or suppressed, harvesting is the primary determinant of forest age-class distributions.

Quantifying forest degradation as a reduction in productivity, carbon stocks or canopy cover also requires that an initial condition (or baseline) is established, against which this reduction is assessed (Section 4.1.4). In forest types with rare stand-replacing disturbances, the concept of ‘intact’ or ‘primary’ forest has been used to define the initial condition (Potapov et al. 2008 <sup>[[#fn:r460|460]]</sup> ) but applying a single metric can be problematic (Bernier et al. 2017 <sup>[[#fn:r461|461]]</sup> ). Moreover, forest types with

frequent stand-replacing disturbances, such as wildfires, or with natural disturbances that reduce carbon stocks, such as some insect outbreaks, experience over time a natural variability of carbon stocks or canopy density, making it more difficult to define the appropriate baseline carbon density or canopy cover against which to assess degradation. In these systems, forest degradation cannot be assessed at the stand level, but requires a landscape-level assessment that takes into consideration the stand age-class distribution of the landscape, which reflects natural and human disturbance regimes over past decades to centuries and also considers post-disturbance regrowth (van Wagner 1978 <sup>[[#fn:r462|462]]</sup> ; Volkova et al. 2018 <sup>[[#fn:r463|463]]</sup> ; Lorimer and White 2003 <sup>[[#fn:r464|464]]</sup> ).

The lack of a consistent definition of forest degradation also affects the ability to establish estimates of the rates or impacts of forest degradation because the drivers of degradation are not clearly defined (Sasaki and Putz 2009 <sup>[[#fn:r465|465]]</sup> ). Moreover, the literature at times confounds estimates of forest degradation and deforestation (i.e., the conversion of forest to non-forest land uses). Deforestation is a change in land use, while forest degradation is not, although severe forest degradation can ultimately lead to deforestation.

Based on empirical data provided by 46 countries, the drivers for deforestation (due to commercial agriculture) and forest degradation (due to timber extraction and logging) are similar in Africa, Asia and Latin America (Hosonuma et al. 2012 <sup>[[#fn:r466|466]]</sup> ). More recently, global forest disturbance over the period 2001–2015 was attributed to commodity-driven deforestation (27 ± 5%), forestry (26 ± 4%), shifting agriculture (24 ± 3%) and wildfire (23 ± 4%). The remaining 0.6 ± 0.3% was attributed to the expansion of urban centres (Curtis et al. 2018 <sup>[[#fn:r467|467]]</sup> ).

The trends of productivity shown by several remote-sensing studies (see previous section) are largely consistent with mapping of forest cover and change using a 34-year time series of coarse resolution satellite data (NOAA AVHRR) (Song et al. 2018 <sup>[[#fn:r468|468]]</sup> ). This study, based on a thematic classification of satellite data, suggests that (i) global tree canopy cover increased by 2.24 million km <sup>2</sup> between 1982 and 2016 (corresponding to +7.1%) but with regional differences that contribute a net loss in the tropics and a net gain at higher latitudes, and (ii) the fraction of bare ground decreased by 1.16 million km <sup>2</sup> (corresponding to –3.1%), mainly in agricultural regions of Asia (Song et al. 2018 <sup>[[#fn:r469|469]]</sup> ), see Figure 4.5. Other tree or land cover datasets show opposite global net trends (Li et al. 2018b <sup>[[#fn:r470|470]]</sup> ), but high agreement in terms of net losses in the tropics and large net gains in the temperate and boreal zones (Li et al. 2018b <sup>[[#fn:r471|471]]</sup> ; Song et al. 2018 <sup>[[#fn:r472|472]]</sup> ; Hansen et al. 2013 <sup>[[#fn:r473|473]]</sup> ). Differences across global estimates are further discussed in Chapter 1 (Section 1.1.2.3) and Chapter 2.

<div id="section-4-3-2-forest-degradation-block-2"></div>

<span id="figure-4.5"></span>
<!-- START IMG -->
<!-- IMG TITLE -->
'''Figure 4.5'''

<span id="diagrams-showing-latitudinal-profiles-of-land-cover-change-over-the-period-1982-to-2016-based-on-analysis-of-time-series-of-noaa-avhrr-imagerya-tree-canopy-cover-change-δtc-b-short-vegetation-cover-change-δsv-c-bare-ground-cover-change-δbg.-area-statistics-were-calculated-for-every-1-of-latitude-song-et-al.-2018.-source-of"></span>
<!-- IMG CAPTION -->
'''Diagrams showing latitudinal profiles of land cover change over the period 1982 to 2016 based on analysis of time-series of NOAA AVHRR imagery:a) tree canopy cover change (ΔTC); b) short vegetation cover change (ΔSV); c) bare ground cover change (ΔBG). Area statistics were calculated for every 1° of latitude (Song et al. 2018). Source of […]'''
<!-- IMG FILE -->
[[File:3014c8afd31535d3f4be58ed0c5e49c5 Figure-4.5-1024x576.jpg]]

Diagrams showing latitudinal profiles of land cover change over the period 1982 to 2016 based on analysis of time-series of NOAA AVHRR imagery:a) tree canopy cover change (ΔTC); b) short vegetation cover change (ΔSV); c) bare ground cover change (ΔBG). Area statistics were calculated for every 1° of latitude (Song et al. 2018 <sup>[[#fn:r1648|1648]]</sup> ). Source of data: NOAA AVHRR.

<!-- END IMG -->
<div id="section-4-3-2-forest-degradation-block-3"></div>

The changes detected from 1982 to 2016 were primarily linked to direct human action, such as land-use changes (about 60% of the observed changes), but also to indirect effects, such as human-induced climate change (about 40% of the observed changes) (Song et al. 2018 <sup>[[#fn:r474|474]]</sup> ), a finding also supported by a more recent study (Chen et al. 2019 <sup>[[#fn:r475|475]]</sup> ). The climate-induced effects were clearly discernible in some regions, such as forest decline in the US Northwest due to increasing pest infestation and increasing fire frequency (Lesk et al. 2017 <sup>[[#fn:r476|476]]</sup> ; Abatzoglou and Williams 2016 <sup>[[#fn:r477|477]]</sup> ; Seidl et al. 2017 <sup>[[#fn:r478|478]]</sup> ), warming-induced

vegetation increase in the Arctic region, general greening in the Sahel probably as a result of increasing rainfall and atmospheric CO <sub>2</sub> , and advancing treelines in mountain regions (Song et al. 2018 <sup>[[#fn:r479|479]]</sup> ). Keenan et al. (2015) <sup>[[#fn:r480|480]]</sup> and Sloan and Sayer (2015) <sup>[[#fn:r481|481]]</sup> studied the 2015 Forest Resources Assessment (FRA) of the Food and Agriculture Organization of the United Nations (FAO) (FAO 2016 <sup>[[#fn:r482|482]]</sup> ) and found that the total forest area from 1990 to 2015 declined by 3%, an estimate that is supported by a global remote-sensing assessment of forest area change that found a 2.8% decline between 1990–2010 (D’Annunzio et al. 2017 <sup>[[#fn:r483|483]]</sup> ; Lindquist and D’Annunzio 2016 <sup>[[#fn:r484|484]]</sup> ). The trend in deforestation is, however, contradicted between these two global assessments, with FAO (2016) suggesting that deforestation is slowing down, while the remote sensing assessments finds it to be accelerating (D’Annunzio et al. 2017 <sup>[[#fn:r485|485]]</sup> ). Recent estimates (Song et al. 2018 <sup>[[#fn:r486|486]]</sup> ) owing to semantic and methodological differences (see Chapter 1, Section 1.1.2.3) suggest that global tree cover has increased over the period 1982–2016, which contradicts the forest area dynamics assessed by FAO (2016) <sup>[[#fn:r487|487]]</sup> and Lindquist and D’Annunzio (2016) <sup>[[#fn:r488|488]]</sup> . The loss rate in tropical forest areas from 2010 to 2015 is 55,000 km <sup>2</sup> yr <sup>-1</sup> . According to the FRA, the global natural forest area also declined from 39.61 Mkm <sup>2</sup> to 37.21 Mkm <sup>2</sup> during the period 1990 to 2015 (Keenan et al. 2015 <sup>[[#fn:r489|489]]</sup> ).

Since 1850, deforestation globally contributed 77% of the emissions from land-use and land-cover change while degradation contributed 10% (with the remainder originating from non-forest land uses) (Houghton and Nassikas 2018 <sup>[[#fn:r490|490]]</sup> ). That study also showed large temporal and regional differences with northern mid-latitude forests currently contributing to carbon sinks due to increasing forest area and forest management. However, the contribution to carbon emissions of degradation as percentage of total forest emissions (degradation and deforestation) are uncertain, with estimates varying from 25% (Pearson et al. 2017 <sup>[[#fn:r491|491]]</sup> ) to nearly 70% of carbon losses (Baccini et al. 2017 <sup>[[#fn:r492|492]]</sup> ). The 25% estimate refers to an analysis of 74 developing countries within tropical and subtropical regions covering 22 million km <sup>2</sup> for the period 2005–2010, while the 70% estimate refers to an analysis of the tropics for the period 2003–2014, but, by and large, the scope of these studies is the same. Pearson et al. (2017) <sup>[[#fn:r493|493]]</sup> estimated annual gross emissions of 2.1 GtCO <sub>2</sub> , of which 53% were derived from timber harvest, 30% from woodfuel harvest and 17% from forest fire. Estimating gross emissions only, creates a distorted representation of human impacts on the land sector carbon cycle. While forest harvest for timber and fuelwood and land-use change (deforestation) contribute to gross emissions, to quantify impacts on the atmosphere, it is necessary to estimate net emissions, that is, the balance of gross emissions and gross removals of carbon from the atmosphere through forest regrowth (Chazdon et al. 2016a <sup>[[#fn:r494|494]]</sup> ; Poorter et al. 2016 <sup>[[#fn:r495|495]]</sup> ; Sanquetta et al. 2018 <sup>[[#fn:r496|496]]</sup> ).

Current efforts to reduce atmospheric CO <sub>2</sub> concentrations can be supported by reductions in forest-related carbon emissions and increases in sinks, which requires that the net impact of forest management on the atmosphere be evaluated (Griscom et al. 2017 <sup>[[#fn:r497|497]]</sup> ). Forest management and the use of wood products in GHG mitigation strategies result in changes in forest ecosystem carbon stocks, changes in harvested wood product carbon stocks, and potential changes in emissions resulting from the use of wood products and forest biomass that substitute for other emissions-intensive materials such as concrete, steel and fossil fuels (Kurz et al. 2016 <sup>[[#fn:r498|498]]</sup> ; Lemprière et al. 2013 <sup>[[#fn:r499|499]]</sup> ; Nabuurs et al. 2007 <sup>[[#fn:r500|500]]</sup> ). The net impact of these changes on GHG emissions and removals, relative to a scenario without forest mitigation actions, needs to be quantified, (e.g., Werner et al. 2010 <sup>[[#fn:r501|501]]</sup> ; Smyth et al. 2014 <sup>[[#fn:r502|502]]</sup> ; Xu et al. 2018 <sup>[[#fn:r503|503]]</sup> ). Therefore, reductions in forest ecosystem carbon stocks alone are an incomplete estimator of the impacts of forest management on the atmosphere (Nabuurs et al. 2007 <sup>[[#fn:r504|504]]</sup> ; Lemprière et al. 2013 <sup>[[#fn:r505|505]]</sup> ; Kurz et al. 2016 <sup>[[#fn:r506|506]]</sup> ; Chen et al. 2018b <sup>[[#fn:r507|507]]</sup> ). The impacts of forest management and the carbon storage in long-lived products and landfills vary greatly by region, however, because of the typically much shorter lifespan of wood products produced from tropical regions compared to temperate and boreal regions (Earles et al. 2012 <sup>[[#fn:r508|508]]</sup> ; Lewis et al. 2019 <sup>[[#fn:r509|509]]</sup> ; Iordan et al. 2018 <sup>[[#fn:r510|510]]</sup> ) (Section 4.8.4).

Assessments of forest degradation based on remote sensing of changes in canopy density or land cover, (e.g., Hansen et al. 2013 <sup>[[#fn:r511|511]]</sup> ; Pearson et al. 2017 <sup>[[#fn:r512|512]]</sup> ) quantify changes in above-ground biomass carbon stocks and require additional assumptions or model-based analyses to also quantify the impacts on other ecosystem carbon pools including below-ground biomass, litter, woody debris and soil carbon. Depending on the type of disturbance, changes in above-ground biomass may lead to decreases or increases in other carbon pools, for example, windthrow and insect-induced tree mortality may result in losses in above-ground biomass that are (initially) offset by corresponding increases in dead organic matter carbon pools (Yamanoi et al. 2015 <sup>[[#fn:r513|513]]</sup> ; Kurz et al. 2008 <sup>[[#fn:r514|514]]</sup> ), while deforestation will reduce the total ecosystem carbon pool (Houghton et al. 2012 <sup>[[#fn:r515|515]]</sup> ).

A global study of current vegetation carbon stocks (450 Gt C), relative to a hypothetical condition without land use (916 Gt C), attributed 42–47% of carbon stock reductions to land management effects without land-use change, while the remaining 53–58% of carbon stock reductions were attributed to deforestation and other land-use changes (Erb et al. 2018 <sup>[[#fn:r516|516]]</sup> ). While carbon stocks in European forests are lower than hypothetical values in the complete absence of human land use, forest area and carbon stocks have been increasing over recent decades (McGrath et al. 2015 <sup>[[#fn:r517|517]]</sup> ; Kauppi et al. 2018 <sup>[[#fn:r518|518]]</sup> ). Studies by Gingrich et al. (2015) <sup>[[#fn:r519|519]]</sup> on the long-term trends in land use over nine European countries (Albania, Austria, Denmark, Germany, Italy, the Netherlands, Romania, Sweden and the United Kingdom) also show an increase in forest land and reduction in cropland and grazing land from the 19th century to the early 20th century. However, the extent to which human activities have affected the productive capacity of forest lands is poorly understood. Biomass Production Efficiency (BPE), i.e. the fraction of photosynthetic production used for biomass production, was significantly higher in managed forests (0.53) compared to natural forests (0.41) (and it was also higher in managed (0.63) compared to natural (0.44) grasslands) (Campioli et al. 2015 <sup>[[#fn:r521|521]]</sup> ). Managing lands for production may involve trade-offs. For example, a larger proportion of NPP in managed forests is allocated to biomass carbon storage, but lower allocation to fine roots is hypothesised to reduce soil carbon stocks in the long term (Noormets et al. 2015 <sup>[[#fn:r522|522]]</sup> ). Annual volume increment in Finnish forests has more than doubled over the last century, due to increased growing stock, improved forest management and environmental changes (Henttonen et al. 2017 <sup>[[#fn:r523|523]]</sup> ).

As economies evolve, the patterns of land-use and carbon stock changes associated with human expansion into forested areas often include a period of rapid decline of forest area and carbon stocks, recognition of the need for forest conservation and rehabilitation, and a transition to more sustainable land management that is often associated with increasing carbon stocks, (e.g., Birdsey et al. 2006 <sup>[[#fn:r524|524]]</sup> ). Developed and developing countries around the world are in various stages of forest transition (Kauppi et al. 2018 <sup>[[#fn:r525|525]]</sup> ; Meyfroidt and Lambin 2011 <sup>[[#fn:r526|526]]</sup> ). Thus, opportunities exist for SFM to contribute to atmospheric carbon targets through reduction of deforestation and degradation, forest conservation, forest restoration, intensification of management, and enhancements of carbon stocks in forests and harvested wood products (Griscom et al. 2017 <sup>[[#fn:r527|527]]</sup> ) ( ''medium evidence, medium agreement'' ).

<span id="projections-of-land-degradation-in-a-changing-climate"></span>