Land Degradation

Land degradation has accompanied humanity at least since the widespread adoption of agriculture during Neolithic time, some 10,000 to 7,500 years ago (Dotterweich 2013²; Butzer 2005³; Dotterweich 2008⁴) and the associated population increase (Bocquet-Appel 2011⁵). There are indications that the levels of greenhouse gases (GHGs) – particularly carbon dioxide (CO₂) and methane (CH₄) – in the atmosphere already started to increase more than 3,000 years ago as a result of expanding agriculture, clearing of forests, and domestication of wild animals (Fuller et al. 2011⁶; Kaplan et al. 2011⁷; Vavrus et al. 2018⁸; Ellis et al. 2013⁹). While the development of agriculture (cropping and animal husbandry) underpinned the development of civilisations, political institutions and prosperity, farming practices led to conversion of forests and grasslands to farmland, and the heavy reliance on domesticated annual grasses for our food production meant that soils started to deteriorate through seasonal mechanical disturbances (Turner et al. 1990¹⁰; Steffen et al. 2005¹¹; Ojima et al. 1994¹²; Ellis et al. 2013¹³). More recently, urbanisation has significantly altered ecosystems (Cross-Chapter Box 4 in Chapter 2). Since around 1850, about 35% of human-caused CO₂ emissions to the atmosphere has come from land as a combined effect of land degradation and land-use change (Foley et al. 2005¹⁴) and about 38% of the Earth’s land area has been converted to agriculture (Foley et al. 2011¹⁵). See Chapter 2 for more details.

Not all human impacts on land result in degradation according to the definition of land degradation used in this report (Section 4.1.3). There are many examples of long-term sustainably managed land around

the world (such as terraced agricultural systems and sustainably managed forests) although degradation and its management are the focus of this chapter. We also acknowledge that human use of land and ecosystems provides essential goods and services for society (Foley et al. 2005¹⁶; Millennium Ecosystem Assessment 2005¹⁷).

Land degradation was long subject to a polarised scientific debate between disciplines and perspectives in which social scientists often proposed that natural scientists exaggerated land degradation as a global problem (Blaikie and Brookfield 1987¹⁸; Forsyth 1996¹⁹; Lukas 2014²⁰; Zimmerer 1993²¹). The elusiveness of the concept in combination with the difficulties of measuring and monitoring land degradation at global and regional scales by extrapolation and aggregation of empirical studies at local scales, such as the Global Assessment of Soil Degradation database (GLASOD) (Sonneveld and Dent 2009²²) contributed to conflicting views. The conflicting views were not confined to science only, but also caused tension between the scientific understanding of land degradation and policy (Andersson et al. 2011²³; Behnke and Mortimore 2016²⁴; Grainger 2009²⁵; Toulmin and Brock 2016²⁶). Another weakness of many land degradation studies is the exclusion of the views and experiences of the land users, whether farmers or forest-dependent communities (Blaikie and Brookfield 1987²⁷; Fairhead and Scoones 2005²⁸; Warren 2002²⁹; Andersson et al. 2011³⁰). More recently, the polarised views described above have been reconciled under the umbrella of Land Change Science, which has emerged as an interdisciplinary field aimed at examining the dynamics of land cover and land-use as a coupled human-environment system (Turner et al. 2007³¹). A comprehensive discussion about concepts and different perspectives of land degradation was presented in Chapter 2 of the recent report from the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES) on land degradation (Montanarella et al. 2018³²).

In summary, agriculture and clearing of land for food and wood products have been the main drivers of land degradation for millennia (high confidence). This does not mean, however, that agriculture and forestry always cause land degradation (high confidence); sustainable management is possible but not always practised (high confidence). Reasons for this are primarily economic, political and social.

To clarify the scope of this chapter, it is important to start by defining land itself. The Special Report on Climate Change and Land (SRCCL) defines land as ‘the terrestrial portion of the biosphere that comprises the natural resources (soil, near surface air, vegetation and other biota, and water), the ecological processes, topography, and human settlements and infrastructure that operate within that system’ (Henry et al. 2018³³, adapted from FAO 2007³⁴; UNCCD 1994³⁵).

Land degradation is defined in many different ways within the literature, with differing emphases on biodiversity, ecosystem functions and ecosystem services (e.g., Montanarella et al. 2018³⁶). In this report, land degradation is defined as a negative trend in land condition, caused by direct or indirect human-induced processes including anthropogenic climate change, expressed as long-term reduction or loss of at least one of the following: biological productivity, ecological integrity or value to humans. This definition applies to forest and non-forest land: forest degradation is land degradation that occurs in forest land. Soil degradation refers to a subset of land degradation processes that directly affect soil.

Several previous IPCC assessment reports include brief discussions of land degradation. In AR5 WGIII land degradation is one factor contributing to uncertainties of the mitigation potential of land-based ecosystems, particularly in terms of fluxes of soil carbon (Smith et al. 2014, p. 817). In AR5 WGI, soil carbon was discussed comprehensively but not in the context of land degradation, except forest degradation (Ciais et al. 2013⁴⁶) and permafrost degradation (Vaughan et al. 2013⁴⁷). Climate change impacts were discussed comprehensively in AR5 WGII, but land degradation was not prominent. Land-use and land-cover changes were treated comprehensively in terms of effects on the terrestrial carbon stocks and flows (Settele et al. 2015⁴⁸) but links to land degradation were, to a large extent, missing. Land degradation was discussed in relation to human security as one factor which, in combination with extreme weather events, has been proposed to contribute to human migration (Adger et al. 2014⁴⁹), an issue discussed more comprehensively in this chapter (Section 4.7.3). Drivers and processes of degradation by which land-based carbon is released to the atmosphere and/or the long-term reduction in the capacity of the land to remove atmospheric carbon and to store this in biomass and soil carbon, have been discussed in the methodological reports of IPCC (IPCC 2006⁵⁰, 2014a⁵¹) but less so in the assessment reports.

The Special Report on Land Use, Land-Use Change and Forestry (SR-LULUCF) (Watson et al. 2000⁵²) focused on the role of the biosphere in the global cycles of GHG. Land degradation was not addressed in a comprehensive way. Soil erosion was discussed as a process by which soil carbon is lost and the productivity of the land is reduced. Deposition of eroded soil carbon in marine sediments was also mentioned as a possible mechanism for permanent sequestration of terrestrial carbon (Watson et al. 2000, p. 194). The possible impacts of climate change on land productivity and degradation were not discussed comprehensively. Much of the report was about how to account for sources and sinks of terrestrial carbon under the Kyoto Protocol.

SFM is defined using several criteria (see above) and its implementation will typically involve trade-offs among these criteria. The conversion of primary forests to sustainably managed forest ecosystems increases relevant economic, social and other functions but often with adverse impacts on biodiversity (Barlow et al. 2007⁷³). In regions with infrequent or no stand-replacing natural disturbances, the timber yield per hectare harvested in managed secondary forests is typically lower than the yield per hectare from the first harvest in the primary forest (Romero and Putz 2018⁷⁴).

The sustainability of timber yield has been achieved in temperate and boreal forests where intensification of management has resulted in increased growing stocks and increased harvest rates in countries where forests had previously been overexploited (Henttonen et al. 2017⁷⁵; Kauppi et al. 2018⁷⁶). However, intensification of management to increase forest productivity can be associated with reductions in biodiversity. For example, when increased productivity is achieved by periodic thinning and removal of trees that would otherwise die due to competition, thinning reduces the amount of dead organic matter of snags and coarse woody debris that can provide habitat, and this loss reduces biodiversity (Spence 2001⁷⁷; Ehnström 2001⁷⁸) and forest carbon stocks (Russell et al. 2015⁷⁹; Kurz et al. 2013⁸⁰). Recognition of adverse biodiversity impacts of high-yield forestry is leading to modified management aimed at increasing habitat availability through, for example, variable retention logging and continuous cover management (Roberts et al. 2016⁸¹) and through the re-introduction of fire disturbances in landscapes where fires have been suppressed (Allen et al. 2002⁸²). Biodiversity losses are also observed during the transition from primary to managed forests in tropical regions (Barlow et al. 2007⁸³) where tree species diversity can be very high – for example, in the Amazon region, about 16,000 tree species are estimated to exist (ter Steege et al. 2013⁸⁴).

Forest certification schemes have been used to document SFM outcomes (Rametsteiner and Simula 2003⁸⁵) by assessing a set of criteria and indicators (e.g., Lindenmayer et al. 2000⁸⁶). While many of the certified forests are found in temperate and boreal countries (Rametsteiner and Simula 2003⁸⁷; MacDicken et al. 2015⁸⁸), examples from the tropics also show that SFM can improve outcomes. For example, selective logging emits 6% of the tropical GHG annually and improved logging practices can reduce emissions by 44% while maintaining timber production (Ellis et al. 2019⁸⁹). In the Congo Basin, implementing reduced impact logging (RIL-C) practices can cut emissions in half without reducing the timber yield (Umunay et al. 2019⁹⁰). SFM adoption depends on the socio-economic and political context, and its improvement depends mainly on better reporting and verification (Siry et al. 2005⁹¹).

Studies of land and forest degradation are often biased towards biophysical aspects, both in terms of its processes, such as erosion or nutrient depletion, and its observed physical manifestations, such as gullying or low primary productivity. Land users’ own perceptions and knowledge about land conditions and degradation have often been neglected or ignored by both policymakers and scientists (Reed et al. 2007⁹⁵; Forsyth 1996⁹⁶; Andersson et al. 2011⁹⁷). A growing body of work is nevertheless beginning to focus on land degradation through the lens of local land users (Kessler and Stroosnijder 2006⁹⁸; Fairhead and Scoones 2005⁹⁹; Zimmerer 1993¹⁰⁰; Stocking et al. 2001¹⁰¹) and the importance of local and indigenous knowledge within land management is starting to be appreciated (Montanarella et al. 2018¹⁰²). Climate change impacts directly and indirectly on the social reality, the land users, and the ecosystem, and vice versa. Land degradation can also have an impact on climate change (Section 4.6).

The use and management of land is highly gendered and is expected to remain so for the foreseeable future (Kristjanson et al. 2017¹⁰³). Women often have less formal access to land than men and less influence over decisions about land, even if they carry out many of the land management tasks (Jerneck 2018a¹⁰⁴; Elmhirst 2011¹⁰⁵; Toulmin 2009¹⁰⁶; Peters 2004¹⁰⁷; Agarwal 1997¹⁰⁸; Jerneck 2018b¹⁰⁹). Many oft-cited general statements about women’s subordination in agriculture are difficult to substantiate, yet it is clear that gender inequality persists (Doss et al. 2015¹¹⁰). Even if women’s access to land is changing formally (Kumar and Quisumbing 2015¹¹¹), the practical outcome is often limited due to several other factors related to both formal and informal institutional arrangements and values (Lavers 2017¹¹²; Kristjanson et al. 2017¹¹³; Djurfeldt et al. 2018¹¹⁴). Women are also affected differently than men when it comes to climate change, having lower adaptive capacities due to factors such as prevailing land tenure frameworks, less access to other capital assets and dominant cultural practices (Vincent et al. 2014¹¹⁵; Antwi-Agyei et al. 2015¹¹⁶; Gabrielsson et al. 2013¹¹⁷). This affects the options available to women to respond to both land degradation and climate change. Indeed, access to land and other assets (e.g., education and training) is key in shaping land-use and land management strategies (Liu et al. 2018b¹¹⁸; Lambin et al. 2001¹¹⁹). Young people are also often disadvantaged in terms of access to resources and decision-making power, even though they carry out much of the day-to-day work (Wilson et al. 2017¹²⁰; Kosec et al. 2018¹²¹; Naamwintome and Bagson 2013¹²²).

Land rights differ between places and are dependent on the political-economic and legal context (Montanarella et al. 2018¹²³). This means that there is no universally applicable best arrangement. Agriculture in highly erosion-prone regions requires site-specific and long-lasting soil and water conservation measures, such as terraces (Section 4.8.1), which may benefit from secure private land rights (Tarfasa et al. 2018¹²⁴; Soule et al. 2000¹²⁵). Pastoral modes of production and community-based forest management systems are often dominated by, and benefit from, communal land tenure arrangements, which may conflict with agricultural/forestry modernisation policies implying private property rights (Antwi-Agyei et al. 2015¹²⁶; Benjaminsen and Lund 2003¹²⁷; Itkonen 2016¹²⁸; Owour et al. 2011¹²⁹; Gebara 2018¹³⁰).

Cultural ecosystem services, defined as the non-material benefits people obtain from ecosystems through spiritual enrichment, cognitive development, reflection, recreation and aesthetic experiences (Millennium Ecosystem Assessment 2005¹³¹) are closely linked to land and ecosystems, although often under-represented in the literature on ecosystem services (Tengberg et al. 2012¹³²; Hernández-Morcillo et al. 2013¹³³). Climate change interacting with land conditions can impact on cultural aspects, such as sense of place and sense of belonging (Olsson et al. 2014¹³⁴).

A large array of interactive physical, chemical, biological and human processes lead to what we define in this report as land degradation (Johnson and Lewis 2007¹³⁸). The biological productivity, ecological integrity (which encompasses both functional and structural attributes of ecosystems) or the human value (which includes any benefit that people get from the land) of a given territory can deteriorate as the result of processes triggered at scales that range from a single furrow (e.g., water erosion under cultivation) to the landscape level (e.g., salinisation through raising groundwater levels under irrigation). While pressures leading to land degradation are often exerted on specific components of the land systems (i.e., soils, water, biota), once degradation processes start, other components become affected through cascading and interactive effects. For example, different pressures and degradation processes can have convergent effects, as can be the case of overgrazing leading to wind erosion, landscape drainage resulting in wetland drying, and warming causing more frequent burning; all of which can independently lead to reductions of the soil organic matter (SOM) pools as a second-order process. Still, the reduction of organic matter pools is also a first-order process triggered directly by the effects of rising temperatures (Crowther et al. 2016¹³⁹) as well as other climate changes such as precipitation shifts (Viscarra Rossel et al. 2014¹⁴⁰). Beyond this complexity, a practical assessment of the major land degradation processes helps to reveal and categorise the multiple pathways in which climate change exerts a degradation pressure (Table 4.1).

Conversion of freshwater wetlands to agricultural land has historically been a common way of increasing the area of arable land. Despite the small areal extent – about 1% of the earth’s surface (Hu et al. 2017¹⁴¹; Dixon et al. 2016¹⁴²) – freshwater wetlands provide a very large number of ecosystem services, such as groundwater replenishment, flood protection and nutrient retention, and are biodiversity hotspots (Reis et al. 2017¹⁴³; Darrah et al. 2019¹⁴⁴; Montanarella et al. 2018¹⁴⁵). The loss of wetlands since 1900 has been estimated at about 55% globally (Davidson 2014¹⁴⁶) (low confidence) and 35% since 1970 (Darrah et al. 2019¹⁴⁷) (medium confidence) which in many situations pose a problem for adaptation to climate change. Drainage causes loss of wetlands, which can be exacerbated by climate change, further reducing the capacity to adapt to climate change (Barnett et al. 2015¹⁴⁸; Colloff et al. 2016¹⁴⁹; Finlayson et al. 2017¹⁵⁰) (high confidence).

In a review of different approaches and attempts to map global land degradation, Gibbs and Salmon (2015)³³⁵ identified four main approaches to map the global extent of degraded lands: expert opinions (Oldeman and van Lynden 1998³³⁶; Dregne 1998³³⁷; Reed 2005³³⁸; Bot et al. 2000³³⁹); satellite observation of vegetation greenness – for example, remote sensing of Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), Plant Phenology Index (PPI) – (Yengoh et al. 2015³⁴⁰; Bai et al. 2008c³⁴¹; Shi et al. 2017³⁴²; Abdi et al. 2019³⁴³; JRC 2018³⁴⁴); biophysical models (biogeographical/ topological) (Cai et al. 2011b³⁴⁵; Hickler et al. 2005³⁴⁶; Steinkamp and Hickler 2015³⁴⁷; Stoorvogel et al. 2017³⁴⁸); and inventories of land use/ condition. Together they provide a relatively complete evaluation, but none on its own assesses the complexity of the process (Vogt et al. 2011³⁴⁹; Gibbs and Salmon 2015³⁵⁰). There is, however, a robust consensus that remote sensing and field-based methods are critical to assess and monitor land degradation, particularly over large areas (such as global, continental and sub-continental) although there are still knowledge gaps to be filled (Wessels et al. 2007³⁵¹, 2004³⁵²; Prince 2016³⁵³; Ghazoul and Chazdon 2017³⁵⁴) as well as the problem of baseline values (Section 4.1.3).

Remote sensing can provide meaningful proxies of land degradation in terms of severity, temporal development, and areal extent. These proxies of land degradation include several indexes that have been used to assess land conditions, and monitoring changes of land conditions – for example, extent of gullies, severe forms of rill and sheet erosion, and deflation. The presence of open-access, quality controlled and continuously updated global databases of remote sensing data is invaluable, and is the only method for consistent monitoring of large areas over several decades (Sedano et al. 2016³⁵⁵; Brandt et al. 2018b³⁵⁶; Turner 2014³⁵⁷).The NDVI, as a proxy for Net Primary Production (NPP) (see Glossary), is one of the most commonly used methods to assess land degradation, since it indicates land cover, an important factor for soil protection. Although NDVI is not a direct measure of vegetation biomass, there is a close coupling between NDVI integrated over a season and in situ NPP (high agreement, robust evidence) (see Higginbottom et al. 2014³⁵⁸; Andela et al. 2013³⁵⁹; Wessels et al. 2012³⁶⁰).

Distinction between land degradation/improvement and the effects of climate variation is an important and contentious issue (Murthy and Bagchi 2018³⁶¹; Ferner et al. 2018³⁶²).There is no simple and straightforward way to disentangle these two effects. The interaction of different determinants of primary production is not well understood. A key barrier to this is a lack of understanding of the inherent interannual variability of vegetation (Huxman et al. 2004³⁶³; Knapp and Smith 2001³⁶⁴; Ruppert et al. 2012³⁶⁵; Bai et al. 2008a³⁶⁶; Jobbágy and Sala 2000³⁶⁷). One possibility is to compare potential land productivity modelled by vegetation models and actual productivity measured by remote sensing (Seaquist et al. 2009³⁶⁸; Hickler et al. 2005³⁶⁹; van der Esch et al. 2017³⁷⁰), but the difference in spatial resolution, typically 0.5 degrees for vegetation models compared to 0.25–0.5 km for remote sensing data, is hampering the approach. The Moderate Resolution Imaging Spectroradiometer (MODIS) provides higher spatial resolution (up to 0.25 km), delivers data for the EVI, which is calculated in the same way as NDVI, and has showed a robust approach to estimate spatial patterns of global annual primary productivity (Shi et al. 2017³⁷¹; Testa et al. 2018³⁷²).

Another approach to disentangle the effects of climate and land use/ management is to use the Rain Use Efficiency (RUE), defined as the biomass production per unit of rainfall, as an indicator (Le Houerou 1984³⁷³; Prince et al. 1998³⁷⁴; Fensholt et al. 2015³⁷⁵). A variant of the RUE approach is the residual trend (RESTREND) of a NDVI time series, defined as the fraction of the difference between the observed NDVI and the NDVI predicted from climate data (Yengoh et al. 2015³⁷⁶; John et al. 2016³⁷⁷). These two metrics aim to estimate the NPP, rainfall and the time dimensions. They are simple transformations of the same three variables: RUE shows the NPP relationship with rainfall for individual years, while RESTREND is the interannual change of RUE; also, both consider that rainfall is the only variable that affects biomass production. They are legitimate metrics when used appropriately, but in many cases they involve oversimplifications and yield misleading results (Fensholt et al. 2015³⁷⁸; Prince et al. 1998³⁷⁹).

There are no reliable global maps of the extent and severity of land degradation (Gibbs and Salmon 2015⁴²¹; Prince et al. 2018⁴²²; van der Esch et al. 2017⁴²³), despite the fact that land degradation is a severe problem (Turner et al. 2016⁴²⁴). The reasons are both conceptual – that is, how land degradation is defined, using what baseline (Herrick et al. 2019⁴²⁵) or over what time period – and methodological – that is, how it can be measured (Prince et al. 2018⁴²⁶). 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 2002⁴²⁷), 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. 2018⁴²⁸).

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 <10 Mkm2 to 60 Mkm2 (corresponding to 8–45% of the ice-free land area) have been degraded globally (Gibbs and Salmon, 2015⁴²⁹) (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. 2008b⁴³⁰,c⁴³¹) with an update to 2011 (Bai et al. 2015⁴³²). 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 1997⁴³³), 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)⁴³⁴ estimated that at least 3.2 billion people were affected, while Barbier and Hochard (2016⁴³⁵, 2018⁴³⁶) 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. 2015⁴³⁷). 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. 2018⁴³⁸). 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. 2019⁴³⁹).

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. (2015⁴⁴⁰) 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.

Several attempts have been made to map the human footprint on the planet (Čuček et al. 2012⁴⁵⁷; Venter et al. 2016⁴⁵⁸) 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).

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. 2018⁴⁵⁹), 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. 2008⁴⁶⁰) but applying a single metric can be problematic (Bernier et al. 2017⁴⁶¹). 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 1978⁴⁶²; Volkova et al. 2018⁴⁶³; Lorimer and White 2003⁴⁶⁴).

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 2009⁴⁶⁵). 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. 2012⁴⁶⁶). 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. 2018⁴⁶⁷).

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. 2018⁴⁶⁸). This study, based on a thematic classification of satellite data, suggests that (i) global tree canopy cover increased by 2.24 million km² 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 km² (corresponding to –3.1%), mainly in agricultural regions of Asia (Song et al. 2018⁴⁶⁹), see Figure 4.5. Other tree or land cover datasets show opposite global net trends (Li et al. 2018b⁴⁷⁰), but high agreement in terms of net losses in the tropics and large net gains in the temperate and boreal zones (Li et al. 2018b⁴⁷¹; Song et al. 2018⁴⁷²; Hansen et al. 2013⁴⁷³). Differences across global estimates are further discussed in Chapter 1 (Section 1.1.2.3) and Chapter 2.

vegetation increase in the Arctic region, general greening in the Sahel probably as a result of increasing rainfall and atmospheric CO₂, and advancing treelines in mountain regions (Song et al. 2018⁴⁷⁹). Keenan et al. (2015)⁴⁸⁰ and Sloan and Sayer (2015)⁴⁸¹ studied the 2015 Forest Resources Assessment (FRA) of the Food and Agriculture Organization of the United Nations (FAO) (FAO 2016⁴⁸²) 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. 2017⁴⁸³; Lindquist and D’Annunzio 2016⁴⁸⁴). 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. 2017⁴⁸⁵). Recent estimates (Song et al. 2018⁴⁸⁶) 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)⁴⁸⁷ and Lindquist and D’Annunzio (2016)⁴⁸⁸. The loss rate in tropical forest areas from 2010 to 2015 is 55,000 km² yr^-1. According to the FRA, the global natural forest area also declined from 39.61 Mkm² to 37.21 Mkm² during the period 1990 to 2015 (Keenan et al. 2015⁴⁸⁹).

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 2018⁴⁹⁰). 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. 2017⁴⁹¹) to nearly 70% of carbon losses (Baccini et al. 2017⁴⁹²). The 25% estimate refers to an analysis of 74 developing countries within tropical and subtropical regions covering 22 million km² 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)⁴⁹³ estimated annual gross emissions of 2.1 GtCO₂, 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. 2016a⁴⁹⁴; Poorter et al. 2016⁴⁹⁵; Sanquetta et al. 2018⁴⁹⁶).

Current efforts to reduce atmospheric CO₂ 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. 2017⁴⁹⁷). 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. 2016⁴⁹⁸; Lemprière et al. 2013⁴⁹⁹; Nabuurs et al. 2007⁵⁰⁰). 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. 2010⁵⁰¹; Smyth et al. 2014⁵⁰²; Xu et al. 2018⁵⁰³). Therefore, reductions in forest ecosystem carbon stocks alone are an incomplete estimator of the impacts of forest management on the atmosphere (Nabuurs et al. 2007⁵⁰⁴; Lemprière et al. 2013⁵⁰⁵; Kurz et al. 2016⁵⁰⁶; Chen et al. 2018b⁵⁰⁷). 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. 2012⁵⁰⁸; Lewis et al. 2019⁵⁰⁹; Iordan et al. 2018⁵¹⁰) (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. 2013⁵¹¹; Pearson et al. 2017⁵¹²) 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. 2015⁵¹³; Kurz et al. 2008⁵¹⁴), while deforestation will reduce the total ecosystem carbon pool (Houghton et al. 2012⁵¹⁵).

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. 2018⁵¹⁶). 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. 2015⁵¹⁷; Kauppi et al. 2018⁵¹⁸). Studies by Gingrich et al. (2015)⁵¹⁹ 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. 2015⁵²¹). 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. 2015⁵²²). 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. 2017⁵²³).

There are two main levels of uncertainty in assessing the risks of future climate-change-induced land degradation. The first level, where uncertainties are comparatively low, involves changes of the degrading agent, such as erosive power of precipitation, heat stress from increasing temperature extremes (Hüve et al. 2011⁵²⁸), water stress from droughts, and high surface wind speed. The second level of uncertainties, and where the uncertainties are much larger, relates to the above – and below-ground ecological changes as a result of changes in climate, such as rainfall, temperature, and increasing level of CO₂. Vegetation cover is crucial to protect against erosion (Mullan et al. 2012⁵²⁹; García-Ruiz et al. 2015⁵³⁰).

Changes in rainfall patterns, such as distribution in time and space, and intensification of rainfall events will increase the risk of land degradation, both in terms of likelihood and consequences (high agreement, medium evidence). Climate-induced vegetation changes will increase the risk of land degradation in some areas (where vegetation cover will decline) (medium confidence). Landslides are a form of land degradation, induced by extreme rainfall events. There is a strong theoretical reason for increasing landslide activity due to intensification of rainfall, but so far, the empirical evidence that climate change has contributed to landslides is lacking (Crozier 2010¹⁶⁴⁹; Huggel et al. 2012⁵³²; Gariano and Guzzetti 2016⁵³³). Human disturbance may be a more important future trigger than climate change (Froude and Petley 2018⁵³⁴).

Indirect impacts of climate change on land degradation are difficult to quantify because of the many conflating factors. The causes of land-use change are complex, combining physical, biological and socio-economic drivers (Lambin et al. 2001⁶⁶⁶; Lambin and Meyfroidt 2011⁶⁶⁷). One such driver of land-use change is the degradation of agricultural land, which can result in a negative cycle of natural land being converted to agricultural land to sustain production levels. The intensive management of agricultural land can lead to a loss of soil function, negatively impacting on the many ecosystem services provided by soils, including maintenance of water quality and soil carbon sequestration (Smith et al. 2016a⁶⁶⁸). The degradation of soil quality due to cropping is of particular concern in tropical regions, where it results in a loss of productive potential of the land, affecting regional food security and driving conversion of non-agricultural land, such as forestry, to agriculture (Lambin et al. 2003⁶⁶⁹; Drescher et al. 2016⁶⁷⁰; Van der Laan et al. 2017⁶⁷¹). Climate change will exacerbate these negative cycles unless sustainable land management practices are implemented.

Climate change impacts on agricultural productivity (see Chapter 5) will have implications for the intensity of land use and hence exacerbate the risk of increasing land degradation. There will be both localised effects (i.e., climate change impacts on productivity affecting land use in the same region) and teleconnections (i.e., climate change impacts and land-use changes that are spatially and temporally separate) (Wicke et al. 2012⁶⁷²; Pielke et al. 2007⁶⁷³). If global temperature increases beyond 3°C it will have negative yield impacts on all crops (Porter et al. 2014⁶⁷⁴) which, in combination with a doubling of demands by 2050 (Tilman et al. 2011⁶⁷⁵), and increasing competition for land from the expansion of negative emissions technologies (IPCC 2018a⁶⁷⁶; Schleussner et al. 2016⁶⁷⁷), will exert strong pressure on agricultural lands and food security.

In sum, reduced productivity of most agricultural crops will drive land-use changes worldwide (robust evidence, medium agreement), but predicting how this will impact on land degradation is challenging because of several conflating factors. Social change, such as widespread changes in dietary preferences, will have a huge impact on agriculture and hence land degradation (medium evidence, high agreement).

In addition to the traditional land-use drivers (e.g., population growth, agricultural expansion, forest management), a new driver will interact to increase competition for land throughout this century: the potential large-scale implementation of land-based technologies for CO₂ removal (CDR). Land-based CDR includes afforestation and reforestation, bioenergy with carbon capture and storage (BECCS), soil carbon management, biochar and enhanced weathering (Smith et al. 2015⁶⁷⁸; Smith 2016⁶⁷⁹).

The large-scale implementation of high-intensity dedicated energy crops, and harvest of crop and forest residues for bioenergy, could contribute to increases in the area of degraded lands: intensive land management can result in nutrient depletion, over-fertilisation and soil acidification, salinisation (from irrigation without adequate drainage), wet ecosystems drying (from increased evapotranspiration), as well as novel erosion and compaction processes (from high-impact biomass harvesting disturbances) and other land degradation processes described in Section 4.2.1.

Although large-scale implementation of land-based CDR has significant potential risks, the need for negative emissions and the anticipated investments to implement such technologies can also create significant opportunities. Investments into land-based CDR can contribute to halting and reversing land degradation, to the restoration or rehabilitation of degraded and marginal lands (Chazdon and Uriarte 2016⁷⁰³; Fritsche et al. 2017⁷⁰⁴) and can contribute to the goals of LDN (Orr et al. 2017⁷⁰⁵).

Estimates of the global area of degraded land range from less than 10 to 60 Mkm2 (Gibbs and Salmon 2015⁷⁰⁶) (Section 4.3.1). Additionally, large areas are classified as marginal lands and may be suitable for the implementation of bioenergy and land-based CDR (Woods et al. 2015⁷⁰⁷). The yield per hectare of marginal and degraded lands is lower than on fertile lands, and if CDR will be implemented on marginal and degraded lands, this will increase the area demand and costs per unit area of achieving negative emissions (Fritsche et al. 2017⁷⁰⁸). The selection of lands suitable for CDR must be considered carefully to reduce conflicts with existing users, to assess the possible trade-offs in biodiversity contributions of the original and the CDR land uses, to quantify the impacts on water budgets, and to ensure sustainability of the CDR land use.

Land use and land condition prior to the implementation of CDR affect climate change benefits (Harper et al. 2018⁷⁰⁹). Afforestation/ reforestation on degraded lands can increase carbon stocks in vegetation and soil, increase carbon sinks (Amichev et al. 2012⁷¹⁰), and deliver co-benefits for biodiversity and ecosystem services, particularly if a diversity of local species are used. Afforestation and reforestation on native grasslands can reduce soil carbon stocks, although the loss is typically more than compensated by increases in biomass and dead organic matter carbon stocks (Bárcena et al. 2014⁷¹¹; Li et al. 2012⁷¹²; Ovalle-Rivera et al. 2015⁷¹³; Shi et al. 2013⁷¹⁴), and may impact on biodiversity (Li et al. 2012⁷¹⁵).

Strategic incorporation of energy crops into agricultural production systems, applying an integrated landscape management approach, can provide co-benefits for management of land degradation and other environmental objectives. For example, buffers of Miscanthus and other grasses can enhance soil carbon and reduce water pollution (Cacho et al. 2018⁷¹⁶; Odgaard et al. 2019⁷¹⁷), and strip-planting of short-rotation tree crops can reduce the water table where crops are affected by dryland salinity (Robinson et al. 2006⁷¹⁸). Shifting to perennial grain crops has the potential to combine food production with carbon sequestration at a higher rate than annual grain crops and avoid the trade-off between food production and climate change mitigation (Crews et al. 2018⁷¹⁹; de Olivera et al. 2018⁷²⁰; Ryan et al. 2018⁷²¹) (Section 4.9.2).

Changes in land cover can affect surface reflectance, water balances and emissions of volatile organic compounds and thus the non-GHG impacts on the climate system from afforestation/reforestation or planting energy crops (Anderson et al. 2011⁷²²; Bala et al. 2007⁷²³; Betts 2000⁷²⁴; Betts et al. 2007⁷²⁵) (see Section 4.6 for further details). Some of these impacts reinforce the GHG mitigation benefits, while others offset the benefits, with strong local (slope, aspect) and regional (boreal vs. tropical biomes) differences in the outcomes (Li et al. 2015⁷²⁶). Adverse effects on albedo from afforestation with evergreen conifers in boreal zones can be reduced through planting of broadleaf deciduous species (Astrup et al. 2018⁷²⁷; Cai et al. 2011a⁷²⁸; Anderson et al. 2011⁷²⁹).

Combining CDR technologies may prove synergistic. Two soil management techniques with an explicit focus on increasing the soil carbon content rather than promoting soil conservation more broadly have been suggested: addition of biochar to agricultural soils (Section 4.9.5) and addition of ground silicate minerals to soils in order to take up atmospheric CO₂ through chemical weathering (Taylor et al. 2017⁷³⁰; Haque et al. 2019⁷³¹; Beerling 2017⁷³²; Strefler et al. 2018⁷³³). The addition of biochar is comparatively well understood and also field tested at large scale, see Section 4.9.5 for a comprehensive discussion. The addition of silicate minerals to soils is still highly uncertain in terms of its potential (from 95 GtCO₂ yr^–1 (Strefler et al. 2018) to only 2–4 GtCO₂ yr^–1 (Fuss et al. 2018⁷³⁴)) and costs (Schlesinger and Amundson 2018⁷³⁵).

Effectively addressing land degradation through implementation of bioenergy and land-based CDR will require site-specific local knowledge, matching of species with the local land, water balance, nutrient and climatic conditions, ongoing monitoring and, where necessary, adaptation of land management to ensure sustainability under global change (Fritsche et al. 2017⁷³⁶). Effective land governance mechanisms including integrated land-use planning, along with strong sustainability standards could support deployment of energy crops and afforestation/reforestation at appropriate scales and geographical contexts (Fritsche et al. 2017⁷³⁷). Capacity-building and technology transfer through the international cooperation mechanisms of the Paris Agreement could support such efforts. Modelling to inform policy development is most useful when undertaken with close interaction between model developers and other stakeholders including policymakers to ensure that models account for real world constraints (Dooley and Kartha 2018⁷³⁸).

Traditional biomass (fuelwood, charcoal, agricultural residues, animal dung) used for cooking and heating by some 2.8 billion people (38% of global population) in non-OECD countries accounts for more than half of all bioenergy used worldwide (IEA 2017⁷⁴¹; REN21 2018⁷⁴²) (Cross-Chapter Box 7 in Chapter 6). Cooking with traditional biomass has multiple negative impacts on human health, particularly for women, children and youth (Machisa et al. 2013⁷⁴³; Sinha and Ray 2015⁷⁴⁴; Price 2017⁷⁴⁵; Mendum and Njenga 2018⁷⁴⁶; Adefuye et al. 2007⁷⁴⁷) and on household productivity, including high workloads for women and youth (Mendum and Njenga 2018⁷⁴⁸; Brunner et al. 2018⁷⁴⁹; Hou et al. 2018⁷⁵⁰; Njenga et al. 2019⁷⁵¹). Traditional biomass is land-intensive due to reliance on open fires, inefficient stoves and overharvesting of woodfuel, contributing to land degradation, losses in biodiversity and reduced ecosystem services (IEA 2017⁷⁵²; Bailis et al. 2015⁷⁵³; Masera et al. 2015⁷⁵⁴; Specht et al. 2015⁷⁵⁵; Fritsche et al. 2017⁷⁵⁶; Fuso Nerini et al. 2017⁷⁵⁷). Traditional woodfuels account for 1.9–2.3% of global GHG emissions, particularly in ‘hotspots’ of land degradation and fuelwood depletion in eastern Africa and South Asia, such that one-third of traditional woodfuels globally are harvested unsustainably (Bailis et al. 2015⁷⁵⁸). Scenarios to significantly reduce reliance on traditional biomass in developing countries present multiple co-benefits (high evidence, high agreement), including reduced emissions of black carbon, a short-lived climate forcer that also causes respiratory disease (Shindell et al. 2012⁷⁵⁹).

A shift from traditional to modern bioenergy, especially in the African context, contributes to improved livelihoods and can reduce land degradation and impacts on ecosystem services (Smeets et al. 2012⁷⁶⁰; Gasparatos et al. 2018⁷⁶¹; Mudombi et al. 2018⁷⁶²). In Sub-Saharan Africa, most countries mention woodfuel in their Nationally Determined Contribution (NDC) but fail to identify transformational processes to make fuelwood a sustainable energy source compatible with improved forest management (Amugune et al. 2017⁷⁶³). In some regions, especially in South and Southeast Asia, a scarcity of woody biomass may lead to excessive removal and use of agricultural wastes and residues, which contributes to poor soil quality and land degradation (Blanco-Canqui and Lal 2009⁷⁶⁴; Mateos et al. 2017⁷⁶⁵).

In Sub-Saharan Africa, forest degradation is widely associated with charcoal production, although in some tropical areas rapid re-growth can offset forest losses (Hoffmann et al. 2017⁷⁶⁶; McNicol et al. 2018⁷⁶⁷). Overharvesting of wood for charcoal contributes to the high rate of deforestation in Sub-Saharan Africa, which is five times the world average, due in part to corruption and weak governance systems (Sulaiman et al. 2017⁷⁶⁸). Charcoal may also be a by-product of forest clearing for agriculture, with charcoal sale providing immediate income when the land is cleared for food crops (Kiruki et al. 2017⁷⁶⁹; Ndegwa et al. 2016⁷⁷⁰). Besides loss of forest carbon stock, a further concern for climate change is methane and black carbon emissions from fuelwood burning and traditional charcoal-making processes (Bond et al. 2013⁷⁷¹; Patange et al. 2015⁷⁷²; Sparrevik et al. 2015⁷⁷³).

A fundamental difficulty in reducing environmental impacts associated with charcoal lies in the small-scale nature of much charcoal production in Sub-Saharan Africa, leading to challenges in regulating its production and trade, which is often informal, and in some cases illegal, but nevertheless widespread since charcoal is the most important urban cooking fuel (Zulu 2010⁷⁷⁴; Zulu and Richardson 2013⁷⁷⁵; Smith et al. 2015⁷⁷⁶; World Bank 2009⁷⁷⁷). Urbanisation combined with population growth has led to continuously increasing charcoal production. Low efficiency of traditional charcoal production results in a four-fold increase in raw woody biomass required and thus much greater biomass harvest (Hojas-Gascon et al. 2016⁷⁷⁸; Smeets et al. 2012⁷⁷⁹). With continuing urbanisation anticipated, increased charcoal production and use will probably contribute to increasing land pressures and increased land degradation, especially in Sub-Saharan Africa (medium evidence, high agreement).

Although it could be possible to source this biomass more sustainably, the ecosystem and health impacts of this increased demand for cooking fuel would be reduced through use of other renewable fuels or, in some cases, non-renewable fuels (LPG), as well as through improved efficiency in end-use and through better resource and supply chain management (Santos et al. 2017⁷⁸⁰; Smeets et al. 2012⁷⁸¹; Hoffmann et al. 2017⁷⁸²). Integrated response options such as agro-forestry (Chapter 6) and good governance mechanisms for forest and agricultural management (Chapter 7) can support the transition to sustainable energy for households and reduce the environmental impacts of traditional biomass.

Land degradation processes with direct impact on soil and terrestrial biota have great relevance in terms of CO₂ exchange with the atmosphere, given the magnitude and activity of these reservoirs in the global carbon cycle. As the most widespread form of soil degradation, erosion detaches the surface soil material, which typically hosts the highest organic carbon stocks, favouring the mineralisation and release as CO₂. Yet complementary processes such as carbon burial may compensate for this effect, making soil erosion a long-term carbon sink (low agreement, limited evidence), (Wang et al. (2017b)⁷⁸³, but see also Chappell et al. (2016)⁷⁸⁴). Precise estimation of the CO₂ released from eroded lands is challenged by the fact that only a fraction of the detached carbon is eventually lost to the atmosphere. It is important to acknowledge that a substantial fraction of the eroded material may preserve its organic carbon load in field conditions. Moreover, carbon sequestration may be favoured through the burial of both the deposited material and the surface of its hosting soil at the deposition location (Quinton et al. 2010⁷⁸⁵). The cascading effects of erosion on other environmental processes at the affected sites can often cause net CO₂ emissions through their indirect influence on soil fertility, and the balance of organic carbon inputs and outputs, interacting with other non-erosive soil degradation processes (such as nutrient depletion, compaction and salinisation), which can lead to the same net carbon effects (see Table 4.1) (van de Koppel et al. 1997⁷⁸⁶).

As natural and human-induced erosion can result in net carbon storage in very stable buried pools at the deposition locations, degradation in those locations has a high C-release potential. Coastal ecosystems such as mangrove forests, marshes and seagrasses are at typical deposition locations, and their degradation or replacement with other vegetation is resulting in a substantial carbon release (0.15 to 1.02 GtC yr^–1) (Pendleton et al. 2012⁷⁸⁷), which highlights the need for a spatially integrated assessment of land degradation impacts on climate that considers in-situ but also ex-situ emissions.

Cultivation and agricultural management of cultivated land are relevant in terms of global CO₂ land–atmosphere exchange (Section 4.8.1). Besides the initial pulse of CO₂ emissions associated with the onset of cultivation and associated vegetation clearing (Chapter 2), agricultural management practices can increase or reduce carbon losses to the atmosphere. Although global croplands are considered to be at a relatively neutral stage in the current decade (Houghton et al. 2012⁷⁸⁸), this results from a highly uncertain balance between coexisting net losses and gains. Degradation losses of soil and biomass carbon appear to be compensated by gains from soil protection and restoration practices such as cover crops, conservation tillage and nutrient replenishment favouring organic matter build-up. Cover crops, increasingly used to improve soils, have the potential to sequester 0.12 GtC yr^–1 on global croplands with a saturation time of more than 150 years (Poeplau and Don 2015⁷⁸⁹). No-till practices (i.e., tillage elimination favouring crop residue retention in the soil surface) which were implemented to protect soils from erosion and reduce land preparation times, were also seen with optimism as a carbon sequestration option, which today is considered more modest globally and, in some systems, even less certain (VandenBygaart 2016⁷⁹⁹; Cheesman et al. 2016⁷⁹¹; Powlson et al. 2014⁷⁹²). Among soil fertility restoration practices, lime application for acidity correction, increasingly important in tropical regions, can generate a significant net CO₂ source in some soils (Bernoux et al. 2003⁷⁹³; Desalegn et al. 2017⁷⁹⁴).

Land degradation processes in seminatural ecosystems driven by unsustainable uses of their vegetation through logging or grazing lead to reduced plant cover and biomass stocks, causing net carbon releases from soils and plant stocks. Degradation by logging activities is particularly prevalent in developing tropical and subtropical regions, involving carbon releases that exceed by far the biomass of harvested products, including additional vegetation and soil sources that are estimated to reach 0.6 GtC yr^–1 (Pearson et al. 2014, 2017⁷⁹⁵). Excessive grazing pressures pose a more complex picture with variable magnitudes and even signs of carbon exchanges. A general trend of higher carbon losses in humid overgrazed rangelands suggests a high potential for carbon sequestration following the rehabilitation of those systems (Conant and Paustian 2002⁷⁹⁶) with a global potential sequestration of 0.045 GtC yr^-1. A special case of degradation in rangelands is the process leading to the woody encroachment of grass-dominated systems, which can be responsible for declining animal production but high carbon sequestration rates (Asner et al. 2003⁷⁹⁷; Maestre et al. 2009⁷⁹⁸).

Fire regime shifts in wild and seminatural ecosystems can become a degradation process in itself, with high impact on net carbon emission and with underlying interactive human and natural drivers such as burning policies (Van Wilgen et al. 2004¹⁶⁵¹), biological invasions (Brooks et al. 2009⁸⁰⁰), and plant pest/disease spread (Kulakowski et al. 2003⁸⁰¹). Some of these interactive processes affecting unmanaged forests have resulted in massive carbon release, highlighting how degradation feedbacks on climate are not restricted to intensively used land but can affect wild ecosystems as well (Kurz et al. 2008⁸⁰²).

Agricultural land and wetlands represent the dominant source of non-CO₂ greenhouse gases (GHGs) (Chen et al. 2018d⁸⁰³). In agricultural land, the expansion of rice cultivation (increasing CH₄ sources), ruminant stocks and manure disposal (increasing CH₄, N₂O and NH₃ fluxes) and nitrogen over-fertilisation combined with soil acidification (increasing N₂O fluxes) are introducing the major impacts (medium agreement, medium evidence) and their associated emissions appear to be exacerbated by global warming (medium agreement, medium evidence) (Oertel et al. 2016⁸⁰⁴).

As the major sources of global N₂O emissions, over-fertilisation and manure disposal are not only increasing in-situ sources but also stimulating those along the pathway of dissolved inorganic nitrogen transport all the way from draining waters to the ocean (high agreement, medium evidence). Current budgets of anthropogenically fixed nitrogen on the Earth System (Tian et al. 2015⁸⁰⁵; Schaefer et al. 2016⁸⁰⁶; Wang et al. 2017a⁸⁰⁷) suggest that N₂O release from terrestrial soils and wetlands accounts for 10–15% of the emissions, yet many further release fluxes along the hydrological pathway remain uncertain, with emissions from oceanic ‘dead-zones’ being a major aspect of concern (Schlesinger 2009; Rabalais et al. 2014⁸⁰⁸).

Among the physical effects of land degradation, surface albedo changes are those with the most evident impact on the net global radiative balance and net climate warming/cooling. Degradation processes affecting wild and semi-natural ecosystems, such as fire regime changes, woody encroachment, logging and overgrazing, can trigger strong albedo changes before significant biogeochemical shifts take place. In most cases these two types of effects have opposite signs in terms of net radiative forcing, making their joint assessment critical for understanding climate feedbacks (Bright et al. 2015⁸²¹).

In the case of forest degradation or deforestation, the albedo impacts are highly dependent on the latitudinal/climatic belt to which they belong. In boreal forests, the removal or degradation of the tree cover increases albedo (net cooling effect) (medium evidence, high agreement) as the reflective snow cover becomes exposed, which can exceed the net radiative effect of the associated carbon release to the atmosphere (Davin et al. 2010⁸²²; Pinty et al. 2011⁸²³). On the other hand, progressive greening of boreal and temperate forests has contributed to net albedo declines (medium agreement, medium evidence) (Planque et al. 2017⁸²⁴; Li et al. 2018a⁸²⁵). In the northern treeless vegetation belt (tundra), shrub encroachment leads to the opposite effect as the emergence of plant structures above the snow cover level reduce winter-time albedo (Sturm 2005⁸²⁶).

The extent to which albedo shifts can compensate for carbon storage shifts at the global level has not been estimated. A significant but partial compensation takes place in temperate and subtropical dry ecosystems in which radiation levels are higher and carbon stocks smaller compared to their more humid counterparts (medium agreement, medium evidence). In cleared dry woodlands, half of the net global warming effect of net carbon release has been compensated by albedo increase (Houspanossian et al. 2013⁸²⁷), whereas in afforested dry rangelands, albedo declines cancelled one-fifth of the net carbon sequestration (Rotenberg and Yakir 2010⁸²⁸). Other important cases in which albedo effects impose a partial compensation of carbon exchanges are the vegetation shifts associated with wildfires, as shown for the savannahs, shrublands and grasslands of Sub-Saharan Africa (Dintwe et al. 2017⁸²⁹). Besides the net global effects discussed above, albedo shifts can play a significant role in local climate (high agreement, medium evidence), as exemplified by the effect of no-till agriculture reducing local heat extremes in European landscapes (Davin et al. 2014⁸³⁰) and the effects of woody encroachment causing precipitation rises in the North American Great Plains (Ge and Zou 2013⁸³¹). Modelling efforts that integrate ground data from deforested areas worldwide accounting for both physical and biogeochemical effects, indicate that massive global deforestation would have a net warming impact (Lawrence and Vandecar 2015⁸³²) at both local and global levels with highlight non-linear effects of forest loss on climate variables.

Beyond the albedo effects presented above, other physical impacts of land degradation on the atmosphere can contribute to global and regional climate change. Of particular relevance, globally and continentally, are the net cooling effects of dust emissions (low agreement, medium evidence) (Lau and Kim (2007)⁸³³, but see also Huang et al. (2014)⁸³⁴). Anthropogenic emission of mineral particles from degrading land appear to have a similar radiative impact than all other anthropogenic aerosols (Sokolik and Toon 1996⁸³⁵). Dust emissions may explain regional climate anomalies through reinforcing feedbacks, as suggested for the amplification of the intensity, extent and duration of the low precipitation anomaly of the North American Dust Bowl in the 1930s (Cook et al. 2009⁸³⁶). Another source of physical effects on climate are surface roughness changes which, by affecting atmospheric drag, can alter cloud formation and precipitation (low agreement, low evidence), as suggested by modelling studies showing how the massive deployment of solar panels in the Sahara could increase rainfall in the Sahel (Li et al. 2018c⁸³⁷), or how woody encroachment in the Arctic tundra could reduce cloudiness and raise temperature (Cho et al. 2018⁸³⁸). The complex physical effects of deforestation, as explored through modelling, converge into general net regional precipitation declines, tropical temperature increases and boreal temperature declines, while net global effects are less certain (Perugini et al. 2017⁸³⁹). Integrating all the physical effects of land degradation and its recovery or reversal is still a challenge, yet modelling attempts suggest that, over the last three decades, the slow but persistent net global greening caused by the average increase of leaf area in the land has caused a net cooling of the Earth, mainly through the rise in evapotranspiration (Zeng et al. 2017⁸⁴⁰) (low confidence).

This section sets out the relationships between land degradation and poverty, and climate change and poverty, leading to inferences about the three-way links between them. Poverty is multidimensional and includes a lack of access to the whole range of capital assets that can be used to pursue a livelihood. Livelihoods constitute the capabilities, assets and activities that are necessary to make a living (Chambers and Conway 1992⁸⁴⁶; Olsson et al. 2014b⁸⁴⁷).

The literature shows high agreement in terms of speculation that there are potential links between land degradation and poverty. However, studies have not provided robust quantitative assessments of the extent and incidence of poverty within populations affected by land degradation (Barbier and Hochard 2016⁸⁴⁸). Some researchers, for example, Nachtergaele et al. (2011)⁸⁴⁹ estimate that 1.5 billion people were dependent upon degraded land to support their livelihoods in 2007, while >42% of the world’s poor population inhabit degraded areas. However, there is overall low confidence in the evidence base, a lack of studies that look beyond the past and present, and the literature calls for more in-depth research to be undertaken on these issues (Gerber et al. 2014⁸⁵⁰). Recent work by Barbier and Hochard (2018)⁸⁵¹ points to biophysical constraints such as poor soils and limited rainfall, which interact to limit land productivity, suggesting that those farming in climatically less-favourable agricultural areas are challenged by poverty. Studies such as those by Coomes et al. (2011)⁸⁵², focusing on an area in the Amazon, highlight the importance of the initial conditions of land holding in the dominant (shifting) cultivation system in terms of long-term effects on household poverty and future forest cover, showing that initial land tenure and socio-economic aspects can make some areas less favourable too.

Much of the qualitative literature is focused on understanding the livelihood and poverty impacts of degradation through a focus on subsistence agriculture, where farms are small, under traditional or informal tenure and where exposure to environmental (including climate) risks is high (Morton 2007⁸⁵³). In these situations, poorer people lack access to assets (financial, social, human, natural and physical) and in the absence of appropriate institutional supports and social protection, this leaves them sensitive and unable to adapt, so a vicious cycle of poverty and degradation can ensue. To further illustrate the complexity, livelihood assessments often focus on a single snapshot in time. Livelihoods are dynamic and people alter their livelihood activities and strategies depending on the internal and external stressors to which they are responding (O’Brien et al. 2004⁸⁵⁴). When certain livelihood activities and strategies are no longer tenable as a result of land degradation (and may push people into poverty), land degradation can have further effects on issues such as migration (Lee 2009⁸⁵⁵), as people adapt by moving (Section 4.7.3); and may result in conflict (Section 4.7.3), as different groups within society compete for scarce resources, sometimes through non-peaceful actions. Both migration and conflict can lead to land-use changes elsewhere that further fuel climate change through increased emissions.

Similar challenges as for understanding land degradation–poverty linkages are experienced in unravelling the relationship between climate change and poverty. A particular issue in examining climate change–poverty links relates to the common use of aggregate economic statistics like GDP, as the assets and income of the poor constitute a minor proportion of national wealth (Hallegatte et al. 2018⁸⁵⁶). Aggregate quantitative measures also fail to capture the distributions of costs and benefits from climate change. Furthermore, people fall into and out of poverty, with climate change being one of many factors affecting these dynamics, through its impacts on livelihoods. Much of the literature on climate change and poverty tends to look backward rather than forward (Skoufias et al. 2011⁸⁵⁷), providing a snapshot of current or past relationships (for example, Dell et al. (2009)⁸⁵⁸ who examine the relationship between temperature and income (GDP) using cross-sectional data from countries in the Americas). Yet, simulations of future climate change impacts on income or poverty are largely lacking.

Noting the limited evidence that exists that explicitly focuses on the relationship between land degradation, climate change and poverty, Barbier and Hochard (2018b)⁸⁵⁹ suggest that those people living in less-favoured agricultural areas face a poverty–environment trap that can result in increased land degradation under climate change conditions. The emergent relationships between land degradation, climate change and poverty are shown in Figure 4.6 (see also Figure 6.1).

Much of the assessment presented above has considered place-based analyses examining the relationships between poverty, land degradation and climate change in the locations in which these outcomes have occurred. Altieri and Nicholls (2017)⁸⁶⁴ note that, due to the globalised nature of markets and consumption systems, the impacts of changes in crop yields linked to climate-related land degradation (manifest as lower yields) will be far reaching, beyond the sites and livelihoods experiencing degradation. Despite these teleconnections, farmers living in poverty in developing countries will be especially vulnerable due to their exposure, dependence on the environment for income and limited options to engage in other ways to make a living (Rosenzweig and Hillel 1998⁸⁶⁵). In identifying ways in which these interlinkages can be addressed, Scherr (2000)⁸⁶⁶ observes that key actions that can jointly address poverty and environmental improvement often seek to increase access to natural resources, enhance the productivity of the natural resource assets of the poor, and engage stakeholders in addressing public natural resource management issues. In this regard, it is increasingly recognised that those suffering from, and being vulnerable to, land degradation and

poverty need to have a voice and play a role in the development of solutions, especially where the natural resources and livelihood activities they depend on are further threatened by climate change.

How and where we grow food, compared to where and when we need to consume it, is at the crux of issues surrounding land degradation, climate change and food security, especially because more than 75% of the global land surface (excluding Antarctica) faces rain-fed crop production constraints (Fischer et al. 2009⁸⁶⁷), see also Chapter 5. Taken separately, knowledge on land degradation processes and human-induced climate change has attained a great level of maturity. However, their combined effects on food security, notably food supply, remain underappreciated (Webb et al. 2017b⁸⁶⁸), and quantitative information is lacking. Just a few studies have shown how the interactive effects of the aforementioned challenging, interrelated phenomena can impact on crop productivity and hence food security and quality (Karami et al. 2009⁸⁶⁹; Allen et al. 2001⁸⁷⁰; Högy and Fangmeier 2008⁸⁷¹) (low evidence). Along with socio-economic drivers, climate change accelerates land degradation due to its influence on land-use systems (Millennium Ecosystem Assessment 2005⁸⁷²; UNCCD 2017⁸⁷³), potentially leading to a decline in agri-food system productivity, particularly on the supply side. Increases in temperature and changes in precipitation patterns are expected to have impacts on soil quality, including nutrient availability and assimilation (St.Clair and Lynch 2010⁸⁷⁴). Those climate-related changes are expected to have net negative impacts on agricultural productivity, particularly in tropical regions, though the magnitude of impacts depends on the models used. Anticipated supply-side issues linked to land and climate relate to biocapacity factors (including e.g., whether there is enough water to support agriculture); production factors (e.g., chemical pollution of soil and water resources or lack of soil nutrients) and distribution issues (e.g., decreased availability of and/or accessibility to the necessary diversity of quality food where and when it is needed) (Stringer et al. 2011⁸⁷⁵). Climate-sensitive transport infrastructure is also problematic for food security (Islam et al. 2017), and can lead to increased food waste, while poor siting of roads and transport links can lead to soil erosion and forest loss (Xiao et al. 2017⁸⁷⁷), further feeding back into climate change.

Land degradation may trigger competition for scarce natural resources, potentially leading to migration and/or conflict, though, even with medium evidence, there is low agreement in the literature. Linkages between land degradation and migration occur within a larger context of multi-scale interaction of environmental and non-environmental drivers and processes, including resettlement projects, searches for education and/or income, land shortages, political turmoil, and family-related reasons (McLeman 2017⁸⁸⁶; Hermans and Ide 2019⁸⁸⁷). The complex contribution of climate to migration and conflict hampers retrieving any level of confidence on climate-migration and climate-conflict linkages, therefore constituting a major knowledge gap (Cramer et al. 2014⁸⁸⁸; Hoegh-Guldberg et al. 2018⁸⁸⁹).

There is low evidence on the causal linkages between climate change, land degradation processes (other than desertification) and migration. Existing studies on land degradation and migration – particularly in drylands – largely focus on the effect of rainfall variability and drought, and show how migration serves as adaptation strategy (Piguet et al. 2018⁸⁹⁰; McLeman 2017⁸⁹¹; Chapter 3). For example, in the Ethiopian highlands, severe topsoil erosion and forest degradation is a major environmental stressor which is amplified by recurring droughts, with migration being an important household adaptation strategy (Morrissey 2013⁸⁹²). In the humid tropics, land degradation, mainly as a consequence of deforestation, has been a reported reason for people leaving their homes during the Amazonian colonisation (Hecht 1983⁸⁹⁴) but was also observed more recently, for example in Guatemala, where soil degradation was one of the most frequently cited migration pushes (López-Carr 2012⁸⁹⁵) and Kenya, where households respond to low soil quality by sending temporary migrants for additional income generation (Gray 2011⁸⁹⁶). In contrast, in the Andean highlands and the Pacific coastal plain, migration increased with land quality, probably because revenues from additional agricultural production was invested in costly forms of migration (Gray and Bilsborrow 2013⁸⁹⁷). These mixed results illustrate the complex, non-linear relationship of land degradation–migration linkages and suggest that explaining land degradationand migration linkages requires considering a broad range of socio-ecological conditions (McLeman 2017⁸⁹⁸).

In addition to people moving away from an area due to ‘lost’ livelihood activities, climate-related land degradation can also reduce the availability of livelihood safety nets – environmental assets that people use during times of shocks or stress. For example, Barbier (2000)⁸⁹⁹ notes that wetlands in north-east Nigeria around Hadejia–Jama’are floodplain provide dry season pastures for seminomadic herders, agricultural surpluses for Kano and Borno states, groundwater recharge of the Chad formation aquifer and ‘insurance’ resources in times of drought. The floodplain also supports many migratory bird species. As climate change and land degradation combine, delivery of these multiple services can be undermined, particularly as droughts become more widespread, reducing the utility of this wetland environment as a safety net for people and wildlife alike.

Early studies conducted in Africa hint at a significant causal link between land degradation and violent conflict (Homer-Dixon et al. 1993⁹⁰⁰). For example, Percival and Homer-Dixon (1995)⁹⁰¹ identified land degradation as one of the drivers of the crisis in Rwanda in the early 1990s, which allowed radical forces to stoke ethnic rivalries. With respect to the Darfur conflict, some scholars and United Nations Environment Programme (UNEP) concluded that land degradation, together with other environmental stressors, constitute a major security threat for the Sudanese people (Byers and Dragojlovic 2004⁹⁰²; Sachs 2007⁹⁰³; UNEP 2007⁹⁰⁴). Recent studies show low agreement, suggesting that climate change can increase the likelihood of civil violence if certain economic, political and social factors, including low development and weak governance mechanisms, are present (Scheffran et al. 2012⁹⁰⁵; Benjaminsen et al. 2012⁹⁰⁶). In contrast, Raleigh and Urdal (2007)⁹⁰⁷ found in a global study that land degradation is a weak predictor for armed conflict. As such, studies addressing possible linkages between climate change – a key driver of land degradation – and the risks of conflict have yielded contradictory results, and it remains largely unclear whether land degradation resulting from climate change leads to conflict or cooperation (Salehyan 2008⁹⁰⁸; Solomon et al. 2018⁹⁰⁹).

Land degradation–conflict linkages can be bi-directional. Research suggests that households experiencing natural resource degradation often engage in migration for securing livelihoods (Kreamer 2012⁹¹⁰), which potentially triggers land degradation at the destination, leading to conflict there (Kassa et al. 2017⁹¹¹). While this indeed holds true for some cases, it may not for others, given the complexity of processes, contexts and drivers. Where conflict and violence do ensue, it is often as a result of a lack of appreciation for the cultural practices of others.

Concrete actions on the ground to address land degradation are primarily focused on soil and water conservation. In the context of adaptation to climate change, actions relevant for addressing land degradation are sometimes framed as ecosystem-based adaptation (Scarano 2017⁹³⁵) or Nature-Based Solutions (Nesshöver et al. 2017⁹³⁶), and in an agricultural context, agroecology (see Glossary) provides an important frame. The site-specific biophysical and social conditions, including local and indigenous knowledge, are important for successful implementation of concrete actions.

Responses to land degradation generally take the form of agronomic measures (methods related to managing the vegetation cover), soil management (methods related to tillage, nutrient supply), and mechanical methods (methods resulting in durable changes to the landscape) (Morgan 2005a⁹³⁷). Measures may be combined to reinforce benefits to land quality, as well as improving carbon sequestration that supports climate change mitigation. Some measures offer adaptation options and other co-benefits, such as agroforestry, involving planting fruit trees that can support food security in the face of climate change impacts (Reed and Stringer 2016⁹³⁸), or application of compost or biochar that enhances soil water-holding capacity, so increases resilience to drought.

There are important differences in terms of labour and capital requirements for different technologies, and also implications for land tenure arrangements. Agronomic measures and soil management require generally little extra capital input and comprise activities repeated annually, so have no particular implication for land tenure arrangements. Mechanical methods require substantial upfront investments in terms of capital and labour, resulting in long-lasting structural change, requiring more secure land tenure arrangements (Mekuriaw et al. 2018⁹³⁹). Agroforestry is a particularly important strategy for SLM in the context of climate change because of the large potential to sequester carbon in plants and soil and enhance resilience of agricultural systems (Zomer et al. 2016⁹⁴⁰).

In practice, responses are anchored in scientific research, as well as local, indigenous and traditional knowledge and know-how. For example, studies in the Philippines by Camacho et al. (2016) ²⁵examine how traditional integrated watershed management by indigenous people sustain regulating services vital to agricultural productivity, while delivering co-benefits in the form of biodiversity and ecosystem resilience at a landscape scale. Although responses can be site specific and sustainable at a local scale, the multi-scale interplay of drivers and pressures can nevertheless cause practices that have been sustainable for centuries to become less so. Siahaya et al. (2016) ¹⁰²⁶explore the traditional knowledge that has informed rice cultivation in the uplands of East Borneo, grounded in sophisticated shifting cultivation methods (gilir balik) which have been passed on for generations (more than 200 years) in order to maintain local food production. Gilir balik involves temporary cultivation of plots, after which, abandonment takes place as the land user moves to another plot, leaving the natural (forest) vegetation to return. This approach is considered sustainable if it has the support of other subsistence strategies, adapts to and integrates with the local context, and if the carrying capacity of the system is not surpassed (Siahaya et al. 2016¹⁰²⁷). Often gilir balik cultivation involves intercropping of rice with bananas, cassava and other food crops. Once the abandoned plot has been left to recover such that soil fertility is restored, clearance takes place again and the plot is reused for cultivation. Rice cultivation in this way plays an important role in forest management, with several different types of succession forest being found in the study by Siahaya et al. (2016). Nevertheless, interplay of these practices with other pressures (large-scale land acquisitions for oil palm plantation, logging and mining), risk their future sustainability. Use of fire is critical in processes of land clearance, so there are also trade-offs for climate change mitigation, which have been sparsely assessed.

Interest appears to be growing in understanding how indigenous and local knowledge inform land users’ responses to degradation, as scientists engage farmers as experts in processes of knowledge co-production and co-innovation (Oliver et al. 2012¹⁰²⁸; Bitzer and Bijman 2015¹⁰²⁹). This can help to introduce, implement, adapt and promote the use of locally appropriate responses (Schwilch et al. 2011¹⁰³⁰). Indeed, studies strongly agree on the importance of engaging local populations in both sustainable land and forest management. Meta-analyses in tropical regions that examined both forests in protected areas and community-managed forests suggest that deforestation rates are lower, with less variation in deforestation rates presenting in community-managed forests compared to protected forests (Porter-Bolland et al. 2012¹⁰³¹). This suggests that consideration of the social and economic needs of local human populations is vital in preventing forest degradation (Ward et al. 2018¹⁰³²). However, while disciplines such as ethnopedology seek to record and understand how local people perceive, classify and use soil, and draw on that information to inform its management (Barrera-Bassols and Zinck 2003¹⁰³³), links with climate change and its impacts (perceived and actual) are not generally considered.

Improved stewardship of forests through reduction or avoidance of deforestation and forest degradation, and enhancement of forest carbon stocks can all contribute to land-based natural climate solutions (Angelsen et al. 2018¹⁰³⁴; Sonwa et al. 2011¹⁰³⁵; Griscom et al. 2017¹⁰³⁶). While estimates of annual emissions from tropical deforestation and forest degradation range widely from 0.5 to 3.5 GtC yr^–1 (Baccini et al. 2017¹⁰³⁷; Houghton et al. 2012¹⁰³⁸; Mitchard 2018¹⁰³⁹; see also Chapter 2), they all indicate the large potential to reduce annual emissions from deforestation and forest degradation. Recent estimates of forest extent for Africa in 1900 may result in downward adjustments of historic deforestation and degradation emission estimates (Aleman et al. 2018¹⁰⁴⁰). Emissions from forest degradation in non-Annex I countries have declined marginally from 1.1 GtCO₂ yr^–1 in 2001–2010 to 1 GtCO₂ yr^–1 in 2011–2015, but the relative emissions from degradation compared to deforestation have increased from a quarter to a third (Federici et al. 2015¹⁰⁴¹). Forest sector activities in developing countries were estimated to represent a technical mitigation potential in 2030 of 9 GtCO₂ (Miles et al. 2015). This was partitioned into reduction of deforestation (3.5 GtCO₂), reduction in degradation and forest management (1.7 GtCO₂) and afforestation and reforestation (3.8 GtCO₂). The economic mitigation potential will be lower than the technical potential (Miles et al. 2015¹⁰⁴²).

Natural regeneration of second-growth forests enhances carbon sinks in the global carbon budget (Chazdon and Uriarte 2016¹⁰⁴³). In Latin America, Chazdon et al. (2016)¹⁰⁴⁴ estimated that, in 2008, second-growth forests (up to 60 years old) covered 2.4 Mkm² of land (28.1% of the total study area). Over 40 years, these lands can potentially accumulate 8.5 GtC in above-ground biomass via low-cost natural regeneration or assisted regeneration, corresponding to a total CO₂ sequestration of 31.1 GtCO₂ (Chazdon et al. 2016b). While above-ground biomass carbon stocks are estimated to be declining in the tropics, they are increasing globally due to increasing stocks in temperate and boreal forests (Liu et al. 2015b), consistent with the observations of a global land sector carbon sink (Le Quéré et al. 2013¹⁰⁴⁵; Keenan et al. 2017¹⁰⁴⁶; Pan et al. 2011).

Moving from technical mitigation potentials (Miles et al. 2015¹⁰⁴⁷) to real reduction of emissions from deforestation and forest degradation required transformational changes (Korhonen-Kurki et al. 2018¹⁰⁴⁸). This transformation can be facilitated by two enabling conditions: the presence of already initiated policy change; or the scarcity of forest resources combined with an absence of any effective forestry framework and policies. These authors and others (Angelsen et al. 2018¹⁰⁴⁹) found that the presence of powerful transformational coalitions of domestic pro-REDD+ (the United Nations Collaborative Programme on Reducing Emissions from Deforestation and Forest Degradation in Developing Countries) political actors combined with strong ownership and leadership, regulations and law enforcement, and performance-based funding, can provide a strong incentive for achieving REDD+ goals.

Implementing schemes such as REDD+ and various projects related to the voluntary carbon market is often regarded as a no-regrets investment (Seymour and Angelsen 2012¹⁰⁵⁰) but the social and ecological implications (including those identified in the Cancun Safeguards) must be carefully considered for REDD+ projects to be socially and ecologically sustainable (Jagger et al. 2015¹⁰⁵¹). In 2018, 34 countries have submitted a REDD+ forest reference level and/ or forest reference emission level to the United Nations Framework Convention on Climate Change (UNFCCC). Of these REDD+ reference levels, 95% included the activity ‘reducing deforestation’ while 34% included the activity ‘reducing forest degradation’ (FAO 2018). Five countries submitted REDD+ results in the technical annex to their Biennial Update Report totalling an emission reduction of 6.3 GtCO₂ between 2006 and 2015 (FAO 2018).

Afforestation is another mitigation activity that increases carbon sequestration (Cross-Chapter Box 2 in Chapter 1). Yet, it requires careful consideration about where to plant trees to achieve potential climatic benefits, given an altering of local albedo and turbulent energy fluxes and increasing night-time land surface temperatures (Peng et al. 2014¹⁰⁵²). A recent hydro-climatic modelling effort has shown that forest cover can account for about 40% of the observed decrease in annual runoff (Buendia et al. 2016¹⁰⁵³). A meta-analysis of afforestation in Northern Europe (Bárcena et al. 2014¹⁰⁵⁴) concluded that significant soil organic carbon sequestration in Northern Europe occurs after afforestation of croplands but not grasslands. Additional sequestration occurs in forest floors and biomass carbon stocks. Successful programmes of large-scale afforestation activities in South Korea and China are discussed in-depth in a special case study (Section 4.9.3).

The potential outcome of efforts to reduce emissions from deforestation and degradation in Indonesia through a 2011 moratorium on concessions to convert primary forests to either timber or palm oil uses was evaluated against rates of emissions over the period 2000 to 2010. The study concluded that less than 7% of emissions would have been avoided had the moratorium been implemented in 2000 because it only curtailed emissions due to a subset of drivers of deforestation and degradation (Busch et al. 2015¹⁰⁵⁵).

While reducing deforestation and forest degradation may directly help to meet mitigation goals, SFM aimed at providing timber, fibre, biomass and non-timber resources can provide long-term livelihood for communities, reduce the risk of forest conversion to non-forest uses (settlement, crops, etc.), and maintain land productivity, thus reducing the risks of land degradation (Putz et al. 2012¹⁰⁷¹; Gideon Neba et al. 2014¹⁰⁷²; Sufo Kankeu et al. 2016¹⁰⁷³; Nitcheu Tchiadje et al. 2016¹⁰⁷⁴; Rossi et al. 2017¹⁰⁷⁵).

Developing SFM strategies aimed at contributing towards negative emissions throughout this century requires an understanding of forest management impacts on ecosystem carbon stocks (including soils), carbon sinks, carbon fluxes in harvested wood, carbon storage in harvested wood products, including landfills and the emission reductions achieved through the use of wood products and bioenergy (Nabuurs et al. 2007¹⁰⁷⁶; Lemprière et al. 2013¹⁰⁷⁷; Kurz et al. 2016¹⁰⁷⁸; Law et al. 2018¹⁰⁷⁹; Nabuurs et al. 2017¹⁰⁸⁰). Transitions from natural to managed forest landscapes can involve a reduction in forest carbon stocks, the magnitude of which depends on the initial landscape conditions, the harvest rotation length relative to the frequency and intensity of natural disturbances, and on the age-dependence of managed and natural disturbances (Harmon et al. 1990¹⁰⁸¹; Kurz et al. 1998¹⁰⁸²). Initial landscape conditions, in particular the age-class distribution and therefore carbon stocks of the landscape, strongly affect the mitigation potential of forest management options (Ter-Mikaelian et al. 2013¹⁰⁸³; Kilpeläinen et al. 2017¹⁰⁸⁴). Landscapes with predominantly mature forests may experience larger reductions in carbon stocks during the transition to managed landscapes (Harmon et al. 1990¹⁰⁸⁵; Kurz et al. 1998¹⁰⁸⁶; Lewis et al. 2019¹⁰⁸⁷). In landscapes with predominantly young or recently disturbed forests, SFM can enhance carbon stocks (Henttonen et al. 2017¹⁰⁸⁸).

Forest growth rates, net primary productivity, and net ecosystem productivity are age-dependent, with maximum rates of CO₂ removal (CDR) from the atmosphere occurring in young to medium-aged forests and declining thereafter (Tang et al. 2014¹⁰⁸⁹). In boreal forest ecosystem, estimation of carbon stocks and carbon fluxes indicate that old growth stands are typically small carbon sinks or carbon sources (Gao et al. 2018¹⁰⁹⁰; Taylor et al. 2014¹⁰⁹¹; Hadden and Grelle 2016¹⁰⁹²). In tropical forests, carbon uptake rates in the first 20 years of forest recovery were 11 times higher than uptake rates in old-growth forests (Poorter et al. 2016¹⁰⁹³). Age-dependent increases in forest carbon stocks and declines in forest carbon sinks mean that landscapes with older forests have accumulated more carbon but their sink strength is diminishing, while landscapes with younger forests contain less carbon but they are removing CO₂ from the atmosphere at a much higher rate (Volkova et al. 2017¹⁰⁹⁴; Poorter et al. 2016¹⁰⁹⁵). The rates of CDR are not just age-related but also controlled by many biophysical factors and human activities (Bernal et al. 2018¹⁰⁹⁶). In ecosystems with uneven-aged, multispecies forests, the relationships between carbon stocks and sinks are more difficult and expensive to quantify.

Whether or not forest harvest and use of biomass is contributing to net reductions of atmospheric carbon depends on carbon losses during and following harvest, rates of forest regrowth, and the use of harvested wood and carbon retention in long-lived or short-lived products, as well as the emission reductions achieved through the substitution of emissions-intensive products with wood products (Lemprière et al. 2013¹⁰⁹⁷; Lundmark et al. 2014¹⁰⁹⁸; Xu et al. 2018b¹⁰⁹⁹; Olguin et al. 2018¹¹⁰⁰; Dugan et al. 2018¹¹⁰¹; Chen et al. 2018b¹¹⁰²; Pingoud et al. 2018¹¹⁰³; Seidl et al. 2007¹¹⁰⁴). Studies that ignore changes in forest carbon stocks (such as some lifecycle analyses that assume no impacts of harvest on forest carbon stocks), ignore changes in wood product pools (Mackey et al. 2013¹¹⁰⁵) or assume long-term steady state (Pingoud et al. 2018¹¹⁰⁶), or ignore changes in emissions from substitution benefits (Mackey et al. 2013¹¹⁰⁷; Lewis et al. 2019¹¹⁰⁸) will arrive at diverging conclusions about the benefits of SFM. Moreover, assessments of climate benefits of any mitigation action must also consider the time dynamics of atmospheric impacts, as some actions will have immediate benefits (e.g., avoided deforestation), while others may not achieve net atmospheric benefits for decades or centuries. For example, the climate benefits of woody biomass use for bioenergy depend on several factors, such as the source and alternate fate of the biomass, the energy type it substitutes, and the rates of regrowth of the harvested forest (Laganière et al. 2017¹¹⁰⁹; Ter-Mikaelian et al. 2014¹¹¹⁰; Smyth et al. 2017¹¹¹¹). Conversion of primary forests in regions of very low stand-replacing disturbances to short-rotation plantations where the harvested wood is used for short-lived products with low displacement factors will increase emissions. In general, greater mitigation benefits are achieved if harvested wood products are used for products with long carbon retention time and high displacement factors.

With increasing forest age, carbon sinks in forests will diminish until harvest or natural disturbances, such as wildfire, remove biomass carbon or release it to the atmosphere (Seidl et al. 2017¹¹¹²). While individual trees can accumulate carbon for centuries (Köhl et al. 2017¹¹¹³), stand-level carbon accumulation rates depend on both tree growth and tree mortality rates (Hember et al. 2016¹¹¹⁴; Lewis et al. 2004¹¹¹⁵). SFM, including harvest and forest regeneration, can help maintain active carbon sinks by maintaining a forest age-class distribution that includes a share of young, actively growing stands (Volkova et al. 2018¹¹¹⁶; Nabuurs et al. 2017¹¹¹⁷). The use of the harvested carbon in either long-lived wood products (e.g., for construction), short-lived wood products (e.g., pulp and paper), or biofuels affects the net carbon balance of the forest sector (Lemprière et al. 2013¹¹¹⁸; Matthews et al. 2018¹¹¹⁹). The use of these wood products can further contribute to GHG emission-reduction goals by avoiding the emissions from the products with higher embodied emissions that have been displaced (Nabuurs et al. 2007¹¹²⁰; Lemprière et al. 2013¹¹²¹). In 2007 the IPCC concluded that ‘[i]n the long term, a sustainable forest management strategy aimed at maintaining or increasing forest carbon stocks, while producing an annual sustained yield of timber, fibre or energy from the forest, will generate the largest sustained mitigation benefit’ (Nabuurs et al. 2007¹¹²²). The apparent trade-offs between maximising forest carbon stocks and maximising ecosystem carbon sinks are at the origin of ongoing debates about optimum management strategies to achieve negative emissions (Keith et al. 2014¹¹²³; Kurz et al. 2016¹¹²⁴; Lundmark et al. 2014¹¹²⁵). SFM, including the intensification of carbon-focused management strategies, can make long-term contributions towards negative emissions if the sustainability of management is assured through appropriate governance, monitoring and enforcement. As specified in the definition of SFM, other criteria such as biodiversity must also be considered when assessing mitigation outcomes (Lecina-Diaz et al. 2018¹¹²⁶). Moreover, the impacts of changes in management on albedo and other non-GHG factors also need to be considered (Luyssaert et al. 2018¹¹²⁷) (Chapter 2). The contribution of SFM for negative emissions is strongly affected by the use of the wood products derived from forest harvest and the time horizon over which the carbon balance is assessed. SFM needs to anticipate the impacts of climate change on future tree growth, mortality and disturbances when designing climate change mitigation and adaptation strategies (Valade et al. 2017¹¹²⁸; Seidl et al. 2017¹¹²⁹).

The 1992 United Nations Conference on Environment and Development (UNCED), also known as the Rio de Janeiro Earth Summit, recognised land degradation as a major challenge to sustainable development, and led to the establishment of the UNCCD, which specifically addressed land degradation in the drylands. The UNCCD emphasises sustainable land use to link poverty reduction on one hand and environmental protection on the other. The two other ‘Rio Conventions’ emerging from the UNCED – the UNFCCC and the Convention on Biological Diversity (CBD) – focus on climate change and biodiversity, respectively. The land has been recognised as an aspect of common interest to the three conventions, and SLM is proposed as a unifying theme for current global efforts on combating land degradation, climate change and loss of biodiversity, as well as facilitating land-based adaptation to climate change and sustainable development.

The Global Environmental Facility (GEF) funds developing countries to undertake activities that meet the goals of the conventions and deliver global environmental benefits. Since 2002, the GEF has invested in projects that support SLM through its Land Degradation Focal Area Strategy, to address land degradation within and beyond the drylands.

Under the UNFCCC, parties have devised National Adaptation Plans (NAPs) that identify medium- and long-term adaptation needs. Parties have also developed their climate change mitigation plans, presented as NDCs. These programmes have the potential of assisting the promotion of SLM. It is understood that the root causes of land degradation and successful adaptation will not be realised until holistic solutions to land management are explored. SLM can help address root causes of low productivity, land degradation, loss of income-generating capacity, as well as contribute to the amelioration of the adverse effects of climate change.

The ‘4 per 1000’ (4p1000) initiative (Soussana et al. 2019¹¹³⁰) launched by France during the UNFCCC COP21 in 2015 aims at capturing CO₂ from the atmosphere through changes to agricultural and forestry practices at a rate that would increase the carbon content of soils by 0.4% per year (Rumpel et al. 2018¹¹³¹). If global soil carbon content increases at this rate in the top 30–40 cm, the annual increase in atmospheric CO₂ would be stopped (Dignac et al. 2017¹¹³²). This is an illustration of how extremely important soils are for addressing climate change. The initiative is based on eight steps: stop carbon loss (priority #1 is peat soils); promote carbon uptake; monitor, report and verify impacts; deploy technology for tracking soil carbon; test strategies for implementation and upscaling; involve communities; coordinate policies; and provide support (Rumpel et al. 2018¹¹³³). Questions remain, however, about the extent that the 4p1000 is achievable as a universal goal (van Groenigen et al. 2017¹¹³⁴; Poulton et al. 2018¹¹³⁵; Schlesinger and Amundson 2018¹¹³⁶).

LDN was introduced by the UNCCD at Rio +20, and adopted at UNCCD COP12 (UNCCD 2016a¹¹³⁷). LDN is defined as ‘a state whereby the amount and quality of land resources necessary to support ecosystem functions and services and enhance food security remain stable or increase within specified temporal and spatial scales and ecosystems’(Cowie et al. 2018¹¹³⁸). Pursuit of LDN requires effort to avoid further net loss of the land-based natural capital relative to a reference state, or baseline. LDN encourages a dual-pronged effort involving SLM to reduce the risk of land degradation, combined with efforts in land restoration and rehabilitation, to maintain or enhance land-based natural capital, and its associated ecosystem services (Orr et al. 2017¹¹³⁹; Cowie et al. 2018¹¹⁴⁰). Planning for LDN involves projecting the expected cumulative impacts of land-use and land management decisions, then counterbalancing anticipated losses with measures to achieve equivalent gains, within individual land types (where land type is defined by land potential). Under the LDN framework developed by UNCCD, three primary indicators are used to assess whether LDN is achieved by 2030: land cover change; net primary productivity; and soil organic carbon (Cowie et al. 2018¹¹⁴¹; Sims et al. 2019¹¹⁴²). Achieving LDN therefore requires integrated landscape management that seeks to optimise land use to meet multiple objectives (ecosystem health, food security, human well-being) (Cohen-Shacham et al. 2016¹¹⁴³). The response hierarchy of Avoid > Reduce > Reverse land degradation articulates the priorities in planning LDN interventions. LDN provides the impetus for widespread adoption of SLM and efforts to restore or rehabilitate land. Through its focus, LDN ultimately provides tremendous potential for mitigation of, and adaptation to, climate change by halting and reversing land degradation and transforming land from a carbon source to a sink. There are strong synergies between the concept of LDN and the NDCs of many countries, with linkages to national climate plans. LDN is also closely related to many Sustainable Development Goals (SDG) in the areas of poverty, food security, environmental protection and sustainable use of natural resources (UNCCD 2016b¹¹⁴⁴). The GEF is supporting countries to set LDN targets and implement their LDN plans through its land degradation focal area, which encourages application of integrated landscape approaches to managing land degradation (GEF 2018¹¹⁴⁵).

Resilience refers to the capacity of interconnected social, economic and ecological systems, such as farming systems, to absorb disturbance (e.g., drought, conflict, market collapse), and respond or reorganise, to maintain their essential function, identity and structure. Resilience can be described as ‘coping capacity’. The disturbance may be a shock – sudden events such as a flood or disease epidemic – or it may be a trend that develops slowly, like a drought or market shift. The shocks and trends anticipated to occur due to climate change are expected to exacerbate risk of land degradation. Therefore, assessing and enhancing resilience to climate change is a critical component of designing SLM strategies.

Resilience as an analytical lens is particularly strong in ecology and related research on natural resource management (Folke et al. 2010¹¹⁷⁵; Quinlan et al. 2016¹¹⁷⁶) while, in the social sciences, the relevance of resilience for studying social and ecological interactions is contested

There is a growing recognition that addressing barriers and designing solutions to complex environmental problems, such as land degradation, requires awareness of the larger system into which the problems and solutions are embedded (Laniak et al. 2013¹²⁰³). An ecosystem approach to sustainable land management (SLM) based on an understanding of land degradation processes has been recommended to separate multiple drivers, pressures and impacts (Kassam et al. 2013¹²⁰⁴), but large uncertainty in model projections of future climate, and associated ecosystem processes (IPCC 2013a¹²⁰⁵) pose additional challenges to the implementation of SLM. As discussed earlier in this chapter, many SLM practices, including technologies and approaches, are available that can increase yields and contribute to closing the yield gap between actual and potential crop or pasture yield, while also enhancing resilience to climate change (Yengoh and Ardö 2014¹²⁰⁶; WOCAT n.d.). However, there are often systemic barriers to adoption and scaling up of SLM practices, especially in developing countries.

Uitto (2016)¹²⁰⁷ identified areas that the GEF, the financial mechanism of the UNCCD, UNFCCC and other multilateral environmental agreements, can address to solve global environmental problems. These include: removal of barriers related to knowledge and information; strategies for implementation of technologies and approaches; and institutional capacity. Strengthening these areas would drive transformational change, leading to behavioural change and broader adoption of sustainable environmental practices. Detailed analysis of barriers as well as strategies, methods and approaches to scale up SLM have been undertaken for GEF programmes in Africa, China and globally (Tengberg and Valencia 2018¹²⁰⁸; Liniger et al. 2011¹²⁰⁹; Tengberg et al. 2016¹²¹⁰). A number of interconnected barriers and bottlenecks to the scaling up of SLM have been identified in this context and are related to:

Adaptive, multi-level and participatory governance of social-ecological systems is considered important for regime shifts and transitions to take place (Wieczorek 2018¹²¹⁶) and essential to secure the capacity of environmental assets to support societal development over longer time periods (Folke et al. 2005¹²¹⁷). There is also recognition that effective environmental policies and programmes need to be informed by a comprehensive understanding of the biophysical, social, and economic components and processes of a system, their complex interactions, and how they respond to different changes (Kelly (Letcher) et al. 2013). But blueprint policies will not work, due to the wide diversity of rules and informal institutions used across sectors and regions of the world, especially in traditional societies (Ostrom 2009¹²¹⁸).

The most effective way of removing barriers to funding of SLM has been mainstreaming of SLM objectives and priorities into relevant policy and development frameworks, and combining SLM best practices with economic incentives for land users. As the short-term costs for establishing and maintaining SLM measures are generally high and constitute a barrier to adoption, land users may need to be compensated for generation of longer-term public goods, such as ecosystem services. Cost-benefit analyses can be conducted on SLM interventions to facilitate such compensations (Liniger et al. 2011¹²¹⁹; Nkonya et al. 2016¹²²⁰; Tengberg et al. 2016¹²²¹). The landscape approach is a means to reconcile competing demands on the land and remove barriers to implementation of SLM (e.g., Sayer et al. 2013¹²²²; Bürgi et al. 2017¹²²³). It involves an increased focus on participatory governance, development of new SLM business models, and innovative funding schemes, including insurance (Shames et al. 2014¹²²⁴). The LDN Fund takes a landscape approach and raises private finance for SLM and promotes market-based instruments, such as PES, certification and carbon trading, that can support scaling up of SLM to improve local livelihoods, sequester carbon and enhance the resilience to climate change (Baumber et al. 2019¹²²⁵).

Over half of the world’s population now lives in towns and cities, a proportion that is predicted to increase to about 70% by the middle of the century (United Nations 2015¹²²⁶). Rapid urbanisation is a severe threat to land and the provision of ecosystem services (Seto et al. 2012¹²²⁷). However, as cities expand, the avoidance of land degradation, or the maintenance/enhancement of ecosystem services is rarely considered in planning processes. Instead, economic development and the need for space for construction is prioritised, which can result in substantial pollution of air and water sources, the degradation of existing agricultural areas and indigenous, natural or semi-natural ecosystems both within and outside of urban areas. For instance, urban areas are characterised by extensive impervious surfaces. Degraded, sealed soils beneath these surfaces do not provide the same quality of water retention as intact soils. Urban landscapes comprising 50–90% impervious surfaces can therefore result in 40–83% of rainfall becoming surface water runoff (Pataki et al. 2011¹²²⁸). With rainfall intensity predicted to increase in many parts of the world under climate change (Royal Society 2016¹²²⁹), increased water runoff is going to get worse. Urbanisation, land degradation and climate change are therefore strongly interlinked, suggesting the need for common solutions (Reed and Stringer 2016¹²³⁰).

There is now a large body of research and application demonstrating the importance of retaining urban green infrastructure (UGI) for the delivery of multiple ecosystem services (DG Environment News Alert Service, 2012; Wentworth, 2017¹²³¹) as an important tool to mitigate and adapt to climate change. UGI can be defined as all green elements within a city, including, but not limited to, retained indigenous ecosystems, parks, public greenspaces, green corridors, street trees, urban forests, urban agriculture, green roofs/walls and private domestic gardens (Tzoulas et al. 2007¹²³²). The definition is usually extended to include ‘blue’ infrastructure, such as rivers, lakes, bioswales and other water drainage features. The related concept of Nature-Based Solutions (defined as: living solutions inspired by, continuously supported by and using nature, which are designed to address various societal challenges in a resource-efficient and adaptable manner and to provide simultaneously economic, social, and environmental benefits) has gained considerable traction within the European Commission as one approach to mainstreaming the importance of UGI (Maes and Jacobs 2017¹²³³; European Union 2015¹²³⁴).

The severe ecological perturbation that is inherent in the conversion of native perennial vegetation to annual crops, and the subsequent high frequency of perturbation required to maintain annual crops, results in at least four forms of soil degradation that will be exacerbated by the effects of climate change (Crews et al. 2016¹²⁵⁹). First, soil erosion is a very serious consequence of annual cropping, with median losses exceeding rates of formation by one to two orders of magnitude in conventionally plowed agroecosystems, and while erosion is reduced with conservation tillage, median losses still exceed formation by several fold (Montgomery 2007¹²⁶⁰). More severe storm intensity associated with climate change is expected to cause even greater losses to wind and water erosion (Nearing et al. 2004¹²⁶¹). Second, the periods of time in which live roots are reduced or altogether absent from soils in annual cropping systems allow for substantial losses of nitrogen from fertilised croplands, averaging 50% globally (Ladha et al. 2005¹²⁶²). This low retention of nitrogen is also expected to worsen with more intense weather events (Bowles et al. 2018¹²⁶³). A third impact of annual cropping is the degradation of soil structure caused by tillage, which can reduce infiltration of precipitation, and increase surface runoff. It is predicted that the percentage of precipitation that infiltrates into agricultural soils will decrease further under climate-change scenarios (Basche and DeLonge 2017¹²⁶⁴; Wuest et al. 2006¹²⁶⁵). The fourth form of soil degradation that results from annual cropping is the reduction of soil organic matter (SOM), a topic of particular relevance to climate change mitigation and adaptation.

Undegraded cropland soils can theoretically hold far more SOM (which is about 58% carbon) than they currently do (Soussana et al. 2006¹²⁶⁶). We know this deficiency because, with few exceptions, comparisons between cropland soils and those of proximate mature native ecosystems commonly show a 40–75% decline in soil carbon attributable to agricultural practices. What happens when native ecosystems are converted to agriculture that induces such significant losses of SOM? Wind and water erosion commonly results in preferential removal of light organic matter fractions that can accumulate on or near the soil surface (Lal 2003¹²⁶⁷). In addition to the effects of erosion, the fundamental practices of growing annual food and fibre crops alters both inputs and outputs of organic matter from most agroecosystems, resulting in net reductions in soil carbon equilibria (Soussana et al. 2006¹²⁶⁸; McLauchlan 2006¹²⁶⁹; Crews et al. 2016¹²⁷⁰). Native vegetation of almost all terrestrial ecosystems is dominated by perennial plants, and the below-ground carbon allocation of these perennials is a key variable in determining formation rates of stable soil organic carbon (SOC) (Jastrow et al. 2007¹²⁷¹; Schmidt et al. 2011¹²⁷²). When perennial vegetation is replaced by annual crops, inputs of root-associated carbon (roots, exudates, mycorrhizae) decline substantially. For example, perennial grassland species allocate around 67% of productivity to roots, whereas annual crops allocate between 13–30% (Saugier 2001¹²⁷³; Johnson et al. 2006¹²⁷⁴).

At the same time, inputs of SOC are reduced in annual cropping systems, and losses are increased because of tillage, compared to native perennial vegetation. Tillage breaks apart soil aggregates which, among other functions, are thought to inhibit soil bacteria, fungi and other microbes from consuming and decomposing SOM (Grandy and Neff 2008¹²⁷⁵). Aggregates reduce microbial access to organic matter by restricting physical access to mineral-stabilised organic compounds as well as reducing oxygen availability (Cotrufo et al. 2015¹²⁷⁶; Lehmann and Kleber 2015¹²⁷⁷). When soil aggregates are broken open with tillage in the conversion of native ecosystems to agriculture, microbial consumption of SOC and subsequent respiration of CO₂ increase dramatically, reducing soil carbon stocks (Grandy and Robertson 2006¹²⁷⁸; Grandy and Neff 2008¹²⁷⁹).

Many management approaches are being evaluated to reduce soil degradation in general, especially by increasing mineral-protected forms of SOC in the world’s croplands (Paustian et al. 2016¹²⁸⁰). The menu of approaches being investigated focuses either on increasing below-ground carbon inputs, usually through increases in total crop productivity, or by decreasing microbial activity, usually through reduced soil disturbance (Crews and Rumsey 2017¹²⁸¹). However, the basic biogeochemistry of terrestrial ecosystems managed for production of annual crops presents serious challenges to achieving the standing stocks of SOC accumulated by native ecosystems that preceded agriculture. A novel new approach that is just starting to receive significant attention is the development of perennial cereal, legume and oilseed crops (Glover et al. 2010¹²⁸²; Baker 2017¹²⁸³).

There are two basic strategies that plant breeders and geneticists are using to develop new perennial grain crop species. The first involves making wide hybrid crosses between existing elite lines of annual crops, such as wheat, sorghum and rice, with related wild perennial species in order to introgress perennialism into the genome of the annual (Cox et al. 2018¹²⁸⁴; Huang et al. 2018¹²⁸⁵; Hayes et al. 2018¹²⁸⁶). The other approach is de novo domestication of wild perennial species that have crop-like traits of interest (DeHaan et al. 2016¹²⁸⁷; DeHaan and Van Tassel 2014¹²⁸⁸). New perennial crop species undergoing de novo domestication include intermediate wheatgrass, a relative of wheat that produces grain known as Kernza (DeHaan et al. 2018¹²⁸⁹; Cattani and Asselin 2018¹²⁹⁰) and Silphium integrifolium, an oilseed crop in the sunflower family (Van Tassel et al. 2017¹²⁹¹). Other grain crops receiving attention for perennialisation include pigeon pea, barley, buckwheat and maize (Batello et al. 2014¹²⁹²; Chen et al. 2018c¹²⁹³) and a number of legume species (Schlautman et al. 2018¹²⁹⁴). In most cases, the seed yields of perennial grain crops under development are well below those of elite modern grain varieties. During the period that it will take for intensive breeding efforts to close the yield and other trait gaps between annual and perennial grains, perennial proto-crops may be used for purposes other than grain, including forage production (Ryan et al. 2018¹²⁹⁵). Perennial rice stands out as a high-yielding exception, as its yields matched those of elite local varieties in the Yunnan Province for six growing seasons over three years (Huang et al. 2018¹²⁹⁶).

Perennial grains hold promises of agricultural practices, which can significantly reduce soil erosion and nutrient leakage while sequestering carbon. When cultivated in mixes with N-fixing species (legumes) such polycultures also reduce the need for external inputs of nitrogen – a large source of GHG from conventional agriculture.

Globally, peatlands cover 3–4% of the Earth’s land area (about 430 Mha) (Xu et al. 2018a¹³⁵⁵) and store 26–44% of estimated global SOC (Moore 2002¹³⁵⁶). They are most abundant in high northern latitudes, covering large areas in North America, Russia and Europe. At lower latitudes, the largest areas of tropical peatlands are located in Indonesia, the Congo Basin and the Amazon Basin in the form of peat swamp forests (Gumbricht et al. 2017¹³⁵⁷; Xu et al. 2018a¹³⁵⁸). It is estimated that, while 80–85% of the global peatland areas is still largely in a natural state, they are such carbon-dense ecosystems that degraded peatlands (0.3% of the terrestrial land) are responsible for a disproportional 5% of global anthropogenic CO₂ emissions – that is, an annual addition of 0.9–3 GtCO₂ to the atmosphere (Dommain et al. 2012¹³⁵⁹; IPCC 2014c¹³⁶⁰).

Peatland degradation is not well quantified globally, but regionally peatland degradation can involve a large percentage of the areas. Land-use change and degradation in tropical peatlands have primarily been quantified in Southeast Asia, where drainage and conversion to plantation crops is the dominant transition (Miettinen et al. 2016¹³⁶¹). Degradation of peat swamps in Peru is also a growing concern and one pilot survey showed that more than 70% of the peat swamps were degraded in one region surveyed (Hergoualc’h et al. 2017a¹³⁶²). Around 65,000 km2 or 10% of the European peatland area has been lost and 44% of the remaining European peatlands are degraded (Joosten, H., Tanneberger 2017¹³⁶³). Large areas of fens have been entirely ‘lost’ or greatly reduced in thickness due to peat wastage (Lamers et al. 2015¹³⁶⁴).

The main drivers of the acceleration of peatland degradation in the 20th century were associated with drainage for agriculture, peat extraction and afforestation related activities (burning, over-grazing, fertilisation) with a variable scale and severity of impact depending on existing resources in the various countries (O’Driscoll et al. 2018¹³⁶⁵; Cobb, A.R. et al. Dommain et al. 2018¹³⁶⁶; Lamers et al. 2015¹³⁶⁷). New drivers include urban development, wind farm construction (Smith et al. 2012¹³⁶⁸), hydroelectric development, tar sands mining and recreational uses (Joosten and Tanneberger 2017¹³⁶⁹). Anthropogenic pressures are now affecting peatlands in previously geographically isolated areas with consequences for global environmental concerns and impacts on local livelihoods (Dargie et al. 2017¹³⁷⁰; Lawson et al. 2015¹³⁷¹; Butler et al. 2009¹³⁷²).

Drained and managed peatlands are GHG-emission hotspots (Swails et al. 2018¹³⁷³; Hergoualc’h et al. 2017a, 2017b¹³⁷⁴; Roman-Cuesta et al. 2016¹³⁷⁵). In most cases, lowering of the water table leads to direct and indirect CO₂ and N₂O emissions to the atmosphere, with rates dependent on a range of factors, including the groundwater level and the water content of surface peat layers, nutrient content, temperature, and vegetation communities. The exception is nutrient-limited boreal peatlands (Minkkinen et al. 2018¹³⁷⁶; Ojanen et al. 2014¹³⁷⁷). Drainage also increases erosion and dissolved organic carbon loss, removing stored carbon into streams as dissolved and particulate organic carbon, which ultimately returns to the atmosphere (Moore et al. 2013¹³⁷⁸; Evans et al. 2016¹³⁷⁹).

In tropical peatlands, oil palm is the most widespread plantation crop and, on average, it emits around 40 tCO₂ ha^–1 yr^–1; Acacia plantations for pulpwood are the second most widespread plantation crop and emit around 73 tCO₂ ha^–1 yr^–1 (Drösler et al. 2013¹³⁸⁰). Other land uses typically emit less than 37 tCO₂ ha^-1yr^-1. Total emissions from peatland drainage in the region are estimated to be between 0.07 and 1.1 GtCO₂ yr^–1 (Houghton and Nassikas 2017¹³⁸¹; Frolking et al. 2011¹³⁸²). Land-use change also affects the fluxes of N₂O and CH₄. Undisturbed tropical peatlands emit about 0.8 MtCH₄ yr^-1 and 0.002 MtN₂O yr^-1, while disturbed peatlands emit 0.1 MtCH₄ yr^–1and 0.2 MtN₂O–N yr^–1 (Frolking et al. 2011¹³⁸³). These N₂O emissions are probably low, as new findings show that emissions from fertilised oil palm can exceed 20 kgN₂O–N ha^–1 yr^–1 (Oktarita et al. 2017¹³⁸⁴).

In the temperate and boreal zones, peatland drainage often leads to emissions in the order of 0.9 to 9.5 tCO₂ ha^–1 y^–1 in forestry plantations and 21 to 29 tCO₂ ha^–1 y^–1 in grasslands and croplands. Nutrient-poor sites often continue to be CO₂ sinks for long periods (e.g., 50 years) following drainage and, in some cases, sinks for atmospheric CH₄, even when drainage ditch emissions are considered (Minkkinen et al. 2018¹³⁸⁵; Ojanen et al. 2014¹³⁸⁶). Undisturbed boreal and temperate peatlands emit about 30 MtCH₄ yr^-1 and 0.02 MtN₂O–N yr^-1, while disturbed peatlands emit 0.1 MtCH₄ yr^–1and 0.2 MtN₂O–N yr^–1 (Frolking et al. 2011¹³⁸⁷).

Fire emissions from tropical peatlands are only a serious issue in Southeast Asia, where they are responsible for 634 (66–4070) MtCO₂ yr^–1 (van der Werf et al. 2017¹³⁸⁸). Much of the variability is linked with the El Niño–Southern Oscillation (ENSO), which produces drought conditions in this region. Anomalously active fire seasons have also been observed in non-drought years and this has been attributed to the increasing effect of high temperatures that dry vegetation out during short dry spells in otherwise normal rainfall years (Fernandes et al. 2017¹³⁸⁹; Gaveau et al. 2014¹³⁹⁰). Fires have significant societal impacts; for example, the 2015 fires caused more than 100,000 additional deaths across Indonesia, Malaysia and Singapore, and this event was more than twice as deadly as the 2006 El Niño event (Koplitz et al. 2016¹³⁹¹).

Peatland degradation in other parts of the world differs from Asia. In Africa, for large peat deposits like those found in the Cuvette Centrale in the Congo Basin or in the Okavango inland delta, the principle threat is changing rainfall regimes due to climate variability and change (Weinzierl et al. 2016¹³⁹²; Dargie et al. 2017¹³⁹³). Expansion of agriculture is not yet a major factor in these regions. In the Western Amazon, extraction of non-timber forest products like the fruits of Mauritia flexuosa (moriche palm) and Suri worms are major sources of degradation that lead to losses of carbon stocks (Hergoualc’h et al. 2017a¹³⁹⁴).

The effects of peatland degradation on livelihoods have not been systematically characterised. In places where plantation crops are driving the conversion of peat swamps, the financial benefits can be considerable. One study in Indonesia found that the net present value of an oil palm plantation is between 3,835 and 9,630 USD per ha to land owners (Butler et al. 2009¹³⁹⁵). High financial returns are creating incentives for the expansion of smallholder production in peatlands. Smallholder plantations extend over 22% of the peatlands in insular Southeast Asia compared to 27% for industrial plantations (Miettinen et al. 2016¹³⁹⁶). In places where income is generated from extraction of marketable products, ecosystem degradation probably has a negative effect on livelihoods. For example, the sale of fruits of M. flexuosa in some parts of the western Amazon constitutes as much as 80% of the winter income of many rural households, but information on trade values and value chains of M. flexuosa is still sparse (Sousa et al. 2018¹³⁹⁷; Virapongse et al. 2017¹³⁹⁸).

Biochar is organic matter that is carbonised by heating in an oxygen-limited environment, and used as a soil amendment. The properties of biochar vary widely, dependent on the feedstock and the conditions of production. Biochar could make a significant contribution to mitigating both land degradation and climate change, simultaneously.

Tropical cyclones are normal disturbances that natural ecosystems have been affected by and recovered from for millennia. Climate models mostly predict decreasing frequency of tropical cyclones, but dramatically increasing intensity of the strongest storms, as well as increasing rainfall rates (Bacmeister et al. 2018¹⁴⁸²; Walsh et al. 2016b¹⁴⁸³). Large amplitude fluctuations in the frequency and intensity complicate both the detection and attribution of tropical cyclones to climate change (Lin and Emanuel 2016b). Yet, the force of high-intensity cyclones has increased and is expected to escalate further due to global climate change (medium agreement, robust evidence) (Knutson et al. 2010¹⁴⁸⁴; Bender et al. 2010¹⁴⁸⁵; Vecchi et al. 2008¹⁴⁸⁶; Bhatia et al. 2018¹⁴⁸⁷; Tu et al. 2018¹⁴⁸⁸; Sobel et al. 2016¹⁴⁸⁹). Tropical cyclone paths are also shifting towards the poles, increasing the area subject to tropical cyclones (Sharmila and Walsh 2018¹⁴⁹⁰; Lin and Emanuel 2016b¹⁴⁹¹). Climate change alone will affect the hydrology of individual wetland ecosystems, mostly through changes in precipitation and temperature regimes with great global variability (Erwin 2009¹⁴⁹²). Over the last seven decades, the speed at which tropical cyclones move has decreased significantly, as expected from theory, exacerbating the damage on local communities from increasing rainfall amounts and high wind speed (Kossin 2018¹⁴⁹³). Tropical cyclones will accelerate changes in coastal forest structure and composition. The heterogeneity of land degradation at coasts that are affected by tropical cyclones can be further enhanced by the interaction of its components (for example, rainfall, wind speed, and direction) with topographic and biological factors (for example, species susceptibility) (Luke et al. 2016¹⁴⁹⁴).

Small Island Developing States (SIDS) are particularly affected by land degradation induced by tropical cyclones; recent examples are Matthew (2016) in the Caribbean, and Pam (2015) and Winston (2016) in the Pacific (Klöck and Nunn 2019¹⁴⁹⁵; Handmer and Nalau 2019¹⁴⁹⁶). Even if the Pacific Ocean has experienced cyclones of unprecedented intensity in recent years, their geomorphological effects may not be unprecedented (Terry and Lau 2018¹⁴⁹⁷).

Cyclone impacts on coastal areas is not restricted to SIDS, but a problem for all low-lying coastal areas (Petzold and Magnan 2019¹⁴⁹⁸). The Sundarbans, one of the world’s largest coastal wetlands, covers about one million hectares between Bangladesh and India. Large areas of the Sundarbans mangroves have been converted into paddy fields over the past two centuries and, more recently, into shrimp farms (Ghosh et al. 2015¹⁴⁹⁹). In 2009, cyclone Aila caused incremental stresses on the socio-economic conditions of the Sundarbans coastal communities through rendering huge areas of land unproductive for a long time (Abdullah et al. 2016¹⁵⁰⁰). The impact of Aila was widespread throughout the Sundarbans mangroves, showing changes between the pre- and post-cyclonic period of 20–50% in the enhanced vegetation index (Dutta et al. 2015¹⁵⁰¹), although the magnitude of the effects of the Sundarbans mangroves derived from climate change is not yet defined (Payo et al. 2016¹⁵⁰²; Loucks et al. 2010¹⁵⁰³; Gopal and Chauhan 2006¹⁵⁰⁴; Ghosh et al. 2015¹⁵⁰⁵; Chaudhuri et al. 2015¹⁵⁰⁶). There is high agreement that the joint effect of climate change and land degradation will be very negative for the area, strongly affecting the environmental services provided by these forests, including the extinction of large mammal species (Loucks et al. 2010¹⁵⁰⁷). The changes in vegetation are mainly due to inundation and erosion (Payo et al. 2016¹⁵⁰⁸).

Tropical cyclone Nargis unexpectedly hit the Ayeyarwady River delta (Myanmar) in 2008 with unprecedented and catastrophic damages to livelihoods, destruction of forests and erosion of fields (Fritz et al. 2009¹⁵⁰⁹) as well as eroding the shoreline 148 m compared with the long-term average (1974–2015) of 0.62 m yr^-1. This is an example of the disastrous effects that changing cyclone paths can have on areas previously not affected by cyclones (Fritz et al. 2010¹⁵¹⁰).

Current environmental changes, including climate change, have caused sea levels to rise worldwide, particularly in tropical and subtropical regions (Fasullo and Nerem 2018¹⁵¹⁷). Combined with scarcity of water in river channels, such rises have been instrumental in the intrusion of highly saline seawater inland, posing a threat to coastal areas and an emerging challenge to land managers and policymakers. Assessing the extent of salinisation due to sea water intrusion at a global scale nevertheless remains challenging. Wicke et al. (2011)¹⁵¹⁸ suggest that across the world, approximately 1.1 Gha of land is affected by salt, with 14% of this categorised as forest, wetland or some other form of protected area. Seawater intrusion is generally caused by (i) increased tidal activity, storm surges, cyclones and sea storms due to changing climate, (ii) heavy groundwater extraction or land-use changes as a result of changes in precipitation, and droughts/floods, (iii) coastal erosion as a result of destruction of mangrove forests and wetlands, (iv) construction of vast irrigation canals and drainage networks leading to low river discharge in the deltaic region; and (v) sea level rise contaminating nearby freshwater aquifers as a result of subsurface intrusion (Uddameri et al. 2014¹⁵¹⁹).

The Indus Delta, located in the south-eastern coast of Pakistan near Karachi in the North Arabian Sea, is one of the six largest estuaries in the world, spanning an area of 600,000 ha. The Indus delta is a clear example of seawater intrusion and land degradation due to local as well as up-country climatic and environmental conditions (Rasul et al. 2012¹⁵²⁰). Salinisation and waterlogging in the up-country areas including provinces of Punjab and Sindh is, however, caused by the irrigation network and over-irrigation (Qureshi 2011¹⁵²¹).

Such degradation takes the form of high soil salinity, inundation and waterlogging, erosion and freshwater contamination. The interannual variability of precipitation with flooding conditions in some years and drought conditions in others has caused variable river flows and sediment runoff below Kotri Barrage (about 200 km upstream of the Indus delta). This has affected hydrological processes in the lower reaches of the river and the delta, contributing to the degradation (Rasul et al. 2012¹⁶⁵⁷).

Coastal degradation – for example, beach erosion, coastal squeeze, and coastal biodiversity loss – as a result of rising sea levels is a major concern for low lying coasts and small islands (high confidence). The contribution of climate change to increased coastal degradation has been well documented in AR5 (Nurse et al. 2014¹⁵⁴⁵; Wong et al. 2014¹⁵⁴⁶) and is further discussed in Section 4.4.1.3 as well as in the IPCC Special Report on the Ocean and Cryosphere in a Changing Climate (SROCC). However, coastal degradation can also be indirectly induced by climate change as the result of adaptation measures that involve changes to the coastal environment, for example, coastal protection measures against increased flooding and erosion due to sea level rise, and storm surges transforming the natural coast to a ‘stabilised’ coastline (Cooper and Pile 2014¹⁵⁴⁷; French 2001¹⁵⁴⁸). Every kind of adaptation response option is context-dependent, and, in fact, sea walls play an important role for adaptation in many places. Nonetheless, there are observed cases where the construction of sea walls can be considered ‘maladaptation’ (Barnett and O’Neill 2010¹⁵⁴⁹; Magnan et al. 2016¹⁶⁵⁹) by leading to increased coastal degradation, such as in the case of small islands where, due to limitations of space, coastal retreat is less of an option than in continental coastal zones. There is emerging literature on the implementation of alternative coastal protection measures and mechanisms on small islands to avoid coastal degradation induced by sea walls (e.g., Mycoo and Chadwick 2012; Sovacool 2012¹⁵⁵¹).