Desertification

Processes of desertification are mechanisms by which drylands are degraded. Desertification consists of both biological and non-biological processes. These processes are classified under broad categories of degradation of physical, chemical and biological properties of terrestrial ecosystems. The number of desertification processes is large and they are extensively covered elsewhere (IPBES 2018a; Lal 2016; Racine 2008; UNCCD 2017). Section 4.2.1 and Tables 4.1 and 4.2 in Chapter 4 highlight those which are particularly relevant for this assessment in terms of their links to climate change and land degradation, including desertification.

Drivers of desertification are factors which trigger desertification processes. Initial studies of desertification during the early-to-mid 20th century attributed it entirely to human activities. In one of the influential publications of that time, Lavauden (1927) stated that: “Desertification is purely artificial. It is only the act of the man…” However, such a uni-causal view of desertification was shown to be invalid (Geist et al. 2004; Reynolds et al. 2007) (Sections 3.1.4.2 and 3.1.4.3). Tables 4.1 and 4.2 in Chapter 4 summarise the drivers, linking them to the specific processes of desertification and land degradation under changing climate.

Erosion refers to removal of soil by the physical forces of water, wind, or often caused by farming activities such as tillage (Ginoux et al. 2012). The global estimates of soil erosion differ significantly, depending on scale, study period and method used (García-Ruiz et al. 2015), ranging from approximately 20 Gt yr–¹ to more than 200 Gt yr–¹ (Boix-Fayos et al. 2006; FAO 2015). There is a significant potential for climate change to increase soil erosion by water, particularly in those regions where precipitation volumes and intensity are projected to increase (Panthou et al. 2014; Nearing et al. 2015). On the other hand, while it is a dominant form of erosion in areas such as West Asia and the Arabian Peninsula (Prakash et al. 2015; Klingmüller et al. 2016), there is limited evidence concerning climate change impacts on wind erosion (Tables 4.1 and 4.2 in Chapter 4, and Section 3.5).

Saline and sodic soils (see Glossary) occur naturally in arid, semi-arid and dry sub-humid regions of the world. Climate change or hydrological change can cause soil salinisation by increasing the mineralised groundwater level. However, secondary salinisation occurs when the concentration of dissolved salts in water and soil is increased by anthropogenic processes, mainly through poorly managed irrigation schemes. The threat of soil and groundwater salinisation induced by sea level rise and seawater intrusion are amplified by climate change (Section 4.9.7).

Global warming is expected to accelerate soil organic carbon (SOC) turnover, since the decomposition of the soil organic matter by microbial activity begins with low soil water availability, but this moisture is insufficient for plant productivity (Austin et al. 2004) (Section 3.4.1.1). SOC is also lost due to soil erosion (Lal 2009); therefore, in some dryland areas leading to SOC decline (Sections 3.3.3 and 3.5.2) and the transfer of carbon (C) from soil to the atmosphere (Lal 2009).

Sea surface temperature (SST) anomalies can drive rainfall changes, with implications for desertification processes. North Atlantic SST anomalies are positively correlated with Sahel rainfall anomalies (Knightetal.2006;Gonzalez-Martinetal.2014;Sheenetal.2017). While the eastern tropical Pacific SST anomalies have a negative correlation with Sahel rainfall (Pomposi et al. 2016), a cooler North Atlantic is related to a drier Sahel, with this relationship enhanced if there is a simultaneous relative warming of the South Atlantic (Hoerling et al. 2006). Huber and Fensholt (2011) explored the relationship between SST anomalies and satellite observed Sahel vegetation dynamics, finding similar relationships but with substantial west–east variations in both the significant SST regions and the vegetation response. Concerning the paleoclimatic evidence on aridification after the early Holocene ‘Green Sahara’ period (11,000 to 5000 years ago), Tierney et al. (2017) indicate that a cooling of the North Atlantic played a role (Collins et al. 2017; Otto-Bliesner et al. 2014; Niedermeyer et al. 2009) similar to that found in modern observations. Besides these SST relationships, aerosols have also been suggested as a potential driver of the Sahel droughts (Rotstayn and Lohmann 2002; Booth et al. 2012; Ackerley et al. 2011). For eastern Africa, both recent droughts and decadal declines have been linked to human-induced warming in the western Pacific (Funk et al. 2018).

The literature on the human drivers of desertification is substantial (e.g., D’Odorico et al. 2013; Sietz et al. 2011; Yan and Cai 2015; Sterk et al. 2016; Varghese and Singh 2016) and there have been several comprehensive reviews and assessments of these drivers very recently (Cherlet et al. 2018; IPBES 2018a; UNCCD 2017). IPBES (2018a) identified cropland expansion, unsustainable land management practices including overgrazing by livestock, urban expansion, infrastructure development, and extractive industries as the main drivers of land degradation. IPBES (2018a) also found that the ultimate driver of land degradation is high and growing consumption of land-based resources, e.g., through deforestation and cropland expansion, escalated by population growth. What is particularly relevant in the context of the present assessment is to evaluate if, how and which human drivers of desertification will be modified by climate change effects.

Growing food demand is driving conversion of forests, rangelands, and woodlands into cropland (Bestelmeyer et al. 2015; D’Odorico et al. 2013). Climate change is projected to reduce crop yields across dryland areas (Sections 3.4.1 and 5.2.2), potentially reducing local production of food and feed. Without research breakthroughs mitigating these productivity losses through higher agricultural productivity, and reducing food waste and loss, meeting the increasing food demands of growing populations will require expansion of cropped areas to more marginal areas (with most prime areas in drylands already being under cultivation) (Lambin 2012; Lambin et al. 2013; Eitelberg et al. 2015; Gutiérrez-Elorza 2006; Kapović Solomun et al. 2018). Borrelli et al. (2017) showed that the primary driver of soil erosion in 2012 was cropland expansion. Although local food demands could also be met by importing from other areas, this would mean increasing the pressure on land in those areas (Lambin and Meyfroidt 2011). The net effects of such global agricultural production shifts on land condition in drylands are not known.

Climate change will exacerbate poverty among some categories of dryland populations (Sections 3.4.2 and 3.5.2). Depending on the context, this impact comes through declines in agricultural productivity, changes in agricultural prices and extreme weather events (Hertel and Lobell 2014; Hallegatte and Rozenberg 2017). There is high confidence that poverty limits both capacities to adapt to climate change and availability of financial resources to invest into SLM (Gerber et al. 2014; Way 2016; Vu et al. 2014) (Sections 3.5.2, 3.6.2 and 3.6.3).

Labour mobility is another key human driver that will interact with climate change. Although strong impacts of climate change on migration in dryland areas are disputed, in some places, it is likely to provide an added incentive to migrate (Section 3.4.2.7). Out-migration will have several contradictory effects on desertification. On one hand, it reduces an immediate pressure on land if it leads to less dependence on land for livelihoods (Chen et al. 2014; Liu et al. 2016a). Moreover, migrant remittances could be used to fund the adoption of SLM practices. Labour mobility from agriculture to non-agricultural sectors could allow land consolidation, gradually leading to mechanisation and agricultural intensification (Wang et al. 2014, 2018). On the other hand, this can increase the costs of labour-intensive SLM practices due to lower availability of rural agricultural labour and/or higher rural wages. Out-migration increases the pressure on land if higher wages that rural migrants earn in urban centres will lead to their higher food consumption. Moreover, migrant remittances could also be used to fund land-use expansion to marginal areas (Taylor et al. 2016; Gray and Bilsborrow 2014). The net effect of these opposite mechanisms varies from place to place (Qin and Liao 2016). There is very little literature evaluating these joint effects of climate change, desertification and labour mobility (Section 7.3.2).

There are also many other institutional, policy and socio-economic drivers of desertification, such as land tenure insecurity, lack of property rights, lack of access to markets, and to rural advisory services, lack of technical knowledge and skills, agricultural price distortions, agricultural support and subsidies contributing to desertification, and lack of economic incentives for SLM (D’Odorico et al. 2013; Geist et al. 2004; Moussa et al. 2016; Mythili and Goedecke 2016; Sow et al. 2016; Tun et al. 2015; García-Ruiz 2010). There is no evidence that these factors will be materially affected by climate change, however, serving as drivers of unsustainable land management practices, they do play a very important role in modulating responses for climate change adaptation and mitigation (Section 3.6.3).

Two broad narratives have historically emerged to describe responses of dryland populations to environmental degradation. The first is ‘desertification syndrome’ which describes the vicious cycle of resource degradation and poverty, whereby dryland populations apply unsustainable agricultural practices leading to desertification, and exacerbating their poverty, which then subsequently further limits their capacities to invest in SLM (MEA 2005; Safriel and Adeel 2008). The alternative paradigm is one of ‘drylands development’, which refers to social and technical ingenuity of dryland populations as a driver of dryland sustainability (MEA 2005; Reynolds et al. 2007; Safriel and Adeel 2008). The major difference between these two frameworks is that the ‘drylands development’ paradigm recognises that human activities are not the sole and/or most important drivers of desertification, but there are interactions of human and climatic drivers within coupled social-ecological systems (Reynolds et al. 2007). This led Behnke and Mortimore (2016), and earlier Swift (1996), to conclude that the concept of desertification as irreversible degradation distorts policy and governance in dryland areas. Mortimore (2016) suggested that instead of externally imposed technical solutions, what is needed is for populations in dryland areas to adapt to this variable environment which they cannot control. All in all, there is high confidence that anthropogenic and climatic drivers interact in complex ways in causing desertification. As discussed in Section 3.2.2, the relative influence of human or climatic drivers on desertification varies from place to place (high confidence) (Bestelmeyer et al. 2018; D’Odorico et al. 2013; Geist and Lambin 2004; Kok et al. 2016; Polley et al. 2013; Ravi et al. 2010; Scholes 2009; Sietz et al. 2017; Sietz et al. 2011).

Complex human–environment interactions, coupled with biophysical, social, economic and political factors unique to any given location, render desertification difficult to map at a global scale (Cherlet et al. 2018). Early attempts to assess desertification focused on expert knowledge in order to obtain global coverage in a cost-effective manner. Expert judgement continues to play an important role because degradation remains a subjective feature whose indicators are different from place to place (Sonneveld and Dent 2007). GLASOD (Global Assessment of Human-induced Soil Degradation) estimated nearly 2000 million hectares (Mha) (15.3% of the total land area) had been degraded by the early 1990s since the mid-20th century. GLASOD was criticised for perceived subjectiveness and exaggeration (Helldén and Tottrup 2008; Sonneveld and Dent 2007). Dregne and Chou (1992) found 3000 Mha in drylands (i.e. about 50% of drylands) were undergoing degradation. Significant improvements have been made through the efforts of WOCAT (World Overview of Conservation Approaches and Technologies), LADA (Land Degradation Assessment in Drylands) and DESIRE (Desertification Mitigation and Remediation of Land) who jointly developed a mapping tool for participatory expert assessment, with which land experts can estimate current area coverage, type and trends of land degradation (Reed et al. 2011).

A simple linear trend in NDVI is an unsuitable measure for dryland degradation for several reasons (Wessels et al. 2012; de Jong et al. 2013; Higginbottom and Symeonakis 2014; Le et al. 2016). NDVI is strongly coupled to precipitation in drylands where precipitation has high inter-annual variability. This means that NDVI trend can be dominated by any precipitation trend and is sensitive to wet or dry periods, particularly if they fall near the beginning or end of the time series. Degradation may only occur during part of the time series, while NDVI is stable or even improving during the rest of the time series. This reduces the strength and representativeness of a linear trend. Other factors such as CO₂ fertilisation also influence the NDVI trend. Various techniques have been proposed to address these issues, including the residual trends (RESTREND) method to account for rainfall variability (Evans and Geerken 2004), time-series break point identification methods to find major shifts in the vegetation trends (de Jong et al. 2013; Verbesselt et al. 2010a), and methods to explicitly account for the effect of CO₂ fertilisation (Le et al. 2016).

Using the RESTREND method, Andela et al. (2013) found that human activity contributed to a mixture of improving and degrading regions in drylands. In some locations these regions differed substantially from those identified using the NDVI trend alone, including an increase in the area being desertified in southern Africa and northern Australia, and a decrease in southeast and western Australia and Mongolia. De Jong et al. (2013) examined the NDVI time series for major shifts in vegetation activity and found that 74% of drylands experienced such a shift between 1981 and 2011. This suggests that monotonic linear trends are unsuitable for accurately capturing the changes that have occurred in the majority of the drylands. Le et al. (2016) explicitly accounted for CO₂ fertilisation effect and found that the extent of degraded areas in the world is 3% larger when compared to the linear NDVI trend.

Besides NDVI, there are many vegetation indices derived from satellite data in the optical and infrared wavelengths. Each of these datasets has been derived to overcome some limitation in existing indices. Studies have compared vegetation indices globally (Zhang et al. 2017) and specifically over drylands (Wu 2014). In general, the data from these vegetation indices are available only since around 2000, while NDVI data is available since 1982. With less than 20 years of data, the trend analysis remains problematic with vegetation indices other than NDVI. However, given the various advantages in terms of resolution and other characteristics, these newer vegetation indices will become more useful in the future as more data accumulates.

A major shortcoming of these studies based on vegetation datasets derived from satellite sensors is that they do not account for changes in vegetation composition, thus leading to inaccuracies in the estimation of the extent of degraded areas in drylands. For example, drylands of eastern Africa currently face growing encroachment of invasive plant species, such as Prosopis juliflora (Ayanu et al. 2015), which constitutes land degradation since it leads to losses in economic productivity of affected areas but appears as a greening in the satellite data. Another case study in central Senegal found degradation manifested through a reduction in species richness despite satellite observed greening (Herrmann and Tappan 2013). A number of efforts to identify changes in vegetation composition from satellites have been made (Brandt et al. 2016a, b; Evans and Geerken 2006; Geerken 2009; Geerken et al. 2005; Verbesselt et al. 2010a, b). These depend on well-identified reference NDVI time series for particular vegetation groupings, can only differentiate vegetation types that have distinct spectral phenology signatures, and require extensive ground observations for validation. A recent alternative approach to differentiating woody from herbaceous vegetation involves the combined use of optical/infrared-based vegetation indices, indicating greenness, with microwave based Vegetation Optical Depth (VOD) which is sensitive to both woody and leafy vegetation components (Andela et al. 2013; Tian et al. 2017).

Vegetation Optical Depth (VOD) has been available since the 1980s. VOD is based on microwave measurements and is related to total above-ground biomass water content. Unlike NDVI, which is only sensitive to green canopy cover, VOD is also sensitive to water in woody parts of the vegetation and hence provides a view of vegetation changes that can be complementary to NDVI. Liu et al. (2013) used VOD trends to investigate biomass changes and found that VOD was closely related to precipitation changes in drylands. To complement their work with NDVI, Andela et al. (2013) also applied the RESTREND method to VOD. By interpreting NDVI and VOD trends together they were able to differentiate changes to the herbaceous and woody components of the biomass. They reported that many dryland regions are experiencing an increase in the woody fraction often associated with shrub encroachment and suggest that this was aided by CO₂ fertilisation.

Biophysical models use global datasets that describe climate patterns and soil groups, combined with observations of land use, to define classes of potential productivity and map general land degradation (Gibbs and Salmon 2015). All biophysical models have their own set of assumptions and limitations that contribute to their overall uncertainty, including: model structure; spatial scale; data requirements (with associated errors); spatial heterogeneities of socio-economic conditions; and agricultural technologies used. Models have been used to estimate the vegetation productivity potential of land (Cai et al. 2011) and to understand the causes of observed vegetation changes. Zhu et al. (2016) used an ensemble of ecosystem models to investigate causes of vegetation changes from 1982–2009, using a factorial simulation approach. They found CO₂ fertilisation to be the dominant effect globally, though climate and land-cover change were the dominant effects in various dryland locations. Borrelli et al. (2017) modelled that about 6.1% of the global land area experienced very high soil erosion rates (exceeding 10 Mg ha−¹yr−¹) in 2012, particularly in South America, Africa, and Asia.

Overall, improved estimation and mapping of areas undergoing desertification are needed. This requires a combination of rapidly expanding sources of remotely sensed data, ground observations and new modelling approaches. This is a critical gap, especially in the context of measuring progress towards achieving the land degradation-neutrality target by 2030 in the framework of SDGs.

While global-scale studies provide information for any region, there are numerous studies that focus on sub-continental scales, providing more in-depth analysis and understanding. Regional and local studies are important to detect location-specific trends in desertification and heterogeneous influences of climate change on desertification. However, these regional and local studies use a wide variety of methodologies, making direct comparisons difficult. For details of the methodologies applied by each study refer to the individual papers.

Africa

It is estimated that 46 of the 54 countries in Africa are vulnerable to desertification, with some already affected (Prăvălie 2016). Moderate or higher severity degradation over recent decades has been identified in many river basins including the Nile (42% of area), Niger (50%), Senegal (51%), Volta (67%), Limpopo (66%) and Lake Chad (26%) (Thiombiano and Tourino-Soto 2007).

The Horn of Africa is getting drier (Damberg and AghaKouchak 2014; Marshall et al. 2012) exacerbating the desertification already occurring (Oroda 2001). The observed decline in vegetation cover is diminishing ecosystem services (Pricope et al. 2013). Based on NDVI residuals, Kenya experienced persistent negative (positive) trends over 21.6% (8.9%) of the country, for the period 1992–2015 (Gichenje and Godinho 2018). Fragmentation of habitats, reduction in the range of livestock grazing, and higher stocking rates are considered to be the main drivers for vegetation structure loss in the rangelands of Kenya (Kihiu 2016; Otuoma et al. 2009).

Despite desertification in the Sahel being a major concern since the 1970s, wetting and greening conditions have been observed in this region over the last three decades (Anyamba and Tucker 2005; Huber et al. 2011; Brandt et al. 2015; Rishmawi et al. 2016; Tian et al. 2016; Leroux et al. 2017; Herrmann et al. 2005; Damberg and AghaKouchak 2014). Cropland areas in the Sahel region of West Africa have doubled since 1975, with settlement area increasing by about 150% (Traore et al. 2014). Thomas and Nigam (2018) found that the Sahara expanded by 10% over the 20th century based on annual rainfall. In Burkina Faso, Dimobe et al. (2015) estimated that from 1984 to 2013, bare soils and agricultural lands increased by 18.8% and 89.7%, respectively, while woodland, gallery forest, tree savannahs, shrub savannahs and water bodies decreased by 18.8%, 19.4%, 4.8%, 45.2% and 31.2%, respectively. In Fakara region in Niger, a 5% annual reduction in herbaceous yield between 1994 and 2006 was largely explained by changes in land use, grazing pressure and soil fertility (Hiernaux et al. 2009). Aladejana et al. (2018) found that between 1986 and 2015, 18.6% of the forest cover around the Owena River basin was lost. For the period 1982–2003, Le et al. (2012) found that 8% of the Volta River basin’s landmass had been degraded, with this increasing to 65% after accounting for the effects of CO₂ (and NOx) fertilisation.

Greening has also been observed in parts of southern Africa but it is relatively weak compared to other regions of the continent (Helldén and Tottrup 2008; Fensholt et al. 2012). However, greening can be accompanied by desertification when factors such as decreasing species richness, changes in species composition and shrub encroachment are observed (Smith et al. 2013; Herrmann and Tappan 2013; Kaptué et al. 2015; Herrmann and Sop 2016; Saha et al. 2015) (Sections 3.1.4 and 3.7.3). In the Okavango river Basin in southern Africa, conversion of land towards higher utilisation intensities, unsustainable agricultural practises and overexploitation of the savanna ecosystems have been observed in recent decades (Weinzierl et al. 2016).

In the arid Algerian High Plateaus, desertification due to both climatic and human causes led to the loss of indigenous plant biodiversity between 1975 and 2006 (Hirche et al. 2011). Ayoub (1998) identified 64 Mha in Sudan as degraded, with the Central North Kordofan state being most affected. However, reforestation measures in the last decade sustained by improved rainfall conditions have led to low-medium regrowth conditions in about 20% of the area (Dawelbait and Morari 2012). In Morocco, areas affected by desertification were predominantly on plains with high population and livestock pressure (del Barrio et al. 2016; Kouba et al. 2018; Lahlaoi et al. 2017). The annual costs of soil degradation were estimated at about 1% of Gross Domestic Product (GDP) in Algeria and Egypt, and about 0.5% in Morocco and Tunisia (Réquier-Desjardins and Bied-Charreton 2006).

Asia

Prăvălie (2016) found that desertification is currently affecting 38 of 48 countries in Asia. The changes in drylands in Asia over the period 1982–2011 were mixed, with some areas experiencing vegetation improvement while others showed reduced vegetation (Miao et al. 2015a). Major river basins undergoing salinisation include: Indo-Gangetic Basin in India (Lal and Stewart 2012), Indus Basin in Pakistan (Aslam and Prathapar 2006), Yellow River Basin in China (Chengrui and Dregne 2001), Yinchuan Plain in China (Zhou et al. 2013), Aral Sea Basin of Central Asia (Cai et al. 2003; Pankova 2016; Qadir et al. 2009).

Helldén and Tottrup (2008) highlighted a greening trend in East Asia between 1982 and 2003. Over the past several decades, air temperature and the rainfall increased in the arid and hyper-arid region of Northwest China (Chen et al. 2015; Wang et al. 2017). Within China, rainfall erosivity has shown a positive trend in dryland areas between 1961 and 2012 (Yang and Lu 2015). While water erosion area in Xinjiang, China, has decreased by 23.2%, erosion considered as severe or intense was still increasing (Zhang et al. 2015). Xue et al. (2017) used remote sensing data covering 1975 to 2015 to show that wind-driven desertified land in northern Shanxi in China had expanded until 2000, before contracting again. Li et al. (2012) used satellite data to identify desertification in Inner Mongolia, China and found a link between policy changes and the locations and extent of human-caused desertification. Several oasis regions in China have seen increases in cropland area, while forests, grasslands and available water resources have decreased (Fu et al. 2017; Muyibul et al. 2018; Xie et al. 2014). Between 1990 and 2011 15.3% of Hogno Khaan nature reserve in central Mongolia was subjected to desertification (Lamchin et al. 2016). Using satellite data Liu et al. (2013) found the area of Mongolia undergoing non-climatic desertification was associated with increases in goat density and wildfire occurrence.

Aerosols can act as a vehicle for the long-range transport of nutrients to oceans (Jickells et al. 2005; Okin et al. 2011) and terrestrial land surfaces (Das et al. 2013). In several locations, notably the Atlantic Ocean, the west of northern Africa, and the Pacific Ocean east of northern China, a considerable amount of mineral dust aerosols, sourced from nearby drylands, reaches the oceans. It was estimated that 60% of dust transported off Africa is deposited in the Atlantic Ocean (Kaufman et al. 2005), while 50% of the dust generated in Asia reaches the Pacific Ocean or further (Uno et al. 2009; Zhang et al. 1997). The Sahara is also a major source of dust for the Mediterranean basin (Varga et al. 2014). The direct effect of atmospheric dust over the ocean was found to be a cooling of the ocean surface (limited evidence, high agreement) (Evan and Mukhopadhyay 2010; Evan et al. 2009) with the tropical North Atlantic mixed layer cooling by over 1°C (Evan et al. 2009).

It has been suggested that dust may act as a source of nutrients for the upper ocean biota, enhancing the biological activity and related carbon sink (medium evidence, low agreement) (Lenes et al. 2001; Shaw et al. 2008; Neuer et al. 2004). The overall response depends on the environmental controls on the ocean biota, the type of aerosols including their chemical constituents, and the chemical environment in which they dissolve (Boyd et al. 2010).

Dust deposited on snow can increase the amount of absorbed solar radiation leading to more rapid melting (Painter et al. 2018), impacting a region’s hydrological cycle (high confidence). Dust deposition on snow and ice has been found in many regions of the globe (e.g., Painter et al. 2018; Kaspari et al. 2014; Qian et al. 2015; Painter et al. 2013), however quantification of the effect globally and estimation of future changes in the extent of this effect remain knowledge gaps.

The Millenium Ecosystem Assessement (2005) proposed four classes of ecosystem services: provisioning, regulating, supporting and cultural services (Cross-Chapter Box 8 in Chapter 6). These ecosystem services in drylands are vulnerable to the impacts of climate change due to high variability in temperature, precipitation and soil fertility (Enfors and Gordon 2008; Mortimore 2005). There is high confidence that desertification processes such as soil erosion, secondary salinisation, and overgrazing have negatively impacted provisioning ecosystem services in drylands, particularly food and fodder production (Majeed and Muhammad 2019; Mirzabaev et al. 2016; Qadir et al. 2009; Van Loo et al. 2017; Tokbergenova et al. 2018) (Section 3.4.2.2). Zika and Erb (2009) reported an estimation of NPP losses between 0.8 and 2.0 GtC yr–¹ due to desertification, comparing the potential NPP and the NPP calculated for the year 2000. In terms of climatic factors, although climatic changes between 1976 and 2016 were found to be favourable for crop yields overall in Russia (Ivanov et al. 2018), yield decreases of up to 40–60% in dryland areas were caused by severe and extensive droughts (Ivanov et al. 2018). Increase in temperature can have a direct impact on animals in the form of increased physiological stress (Rojas-Downing et al. 2017), increased water requirements for drinking and cooling, a decrease in the production of milk, meat and eggs, increased stress during conception and reproduction (Nardone et al. 2010) or an increase in seasonal diseases and epidemics (Thornton et al. 2009; Nardone et al. 2010). Furthermore, changes in temperature can indirectly impact livestock through reducing the productivity and quality of feed crops and forages (Thornton et al. 2009; Polley et al. 2013). On the other hand, fewer days with extreme cold temperatures during winter in the temperate zones are associated with lower livestock mortality. The future projection of impacts on ecosystems is presented in Section 3.5.2.

Plant biodiversity

Over 20% of global plant biodiversity centres are located within drylands (White and Nackoney 2003). Plant species located within these areas are characterised by high genetic diversity within populations (Martínez-Palacios et al. 1999). The plant species within these ecosystems are often highly threatened by climate change and desertification (Millennium Ecosystem Assessment 2005b; Maestre et al. 2012). Increasing aridity exacerbates the risk of extinction of some plant species, especially those that are already threatened due to small populations or restricted habitats (Gitay et al. 2002). Desertification, including through land-use change, already contributed to the loss of biodiversity across drylands (medium confidence) (Newbold et al. 2015; Wilting et al. 2017). For example, species richness decreased from 234 species in 1978 to 95 in 2011 following long periods of drought and human driven degradation on the steppe land of south-western Algeria (Observatoire du Sahara et du Sahel 2013). Similarly, drought and overgrazing led to loss of biodiversity in Pakistan to the point that only drought-adapted species can now survive on the arid rangelands (Akhter and Arshad 2006). Similar trends were observed in desert steppes of Mongolia (Khishigbayar et al. 2015). In contrast, the increase in annual moistening of southern European Russia from the late 1980s to the beginning of the 21st century caused the restoration of steppe vegetation, even under conditions of strong anthropogenic pressure (Ivanov et al. 2018). The seed banks of annual species can often survive over the long term, germinating in wet years, suggesting that these species could be resilient to some aspects of climate change (Vetter et al. 2005). Yet, Hiernaux and Houérou (2006) showed that overgrazing in the Sahel tended to decrease the seed bank of annuals, which could make them vulnerable to climate change over time. Perennial species, considered as the structuring element of the ecosystem, are usually less affected as they have deeper roots, xeromorphic properties and physiological mechanisms that increase drought tolerance (Le Houérou 1996). However, in North Africa, long-term monitoring (1978–2014) has shown that important plant perennial species have also disappeared due to drought (Stipa tenacissima and Artemisia herba alba) (Hirche et al. 2018; Observatoire du Sahara et du Sahel 2013). The aridisation of the climate in the south of Eastern Siberia led to the advance of the steppes to the north and to the corresponding migration of steppe mammal species between 1976 and 2016 (Ivanov et al. 2018). The future projection of impacts on plant biodiversity is presented in Section 3.5.2.

Climate change has a high potential to contribute to poverty particularly through the risks coming from extreme weather events (Olsson et al. 2014). However, the evidence rigourously attributing changes in observed poverty to climate change impacts is currently not available. On the other hand, most of the research on links between poverty and desertification (or more broadly, land degradation) focused on whether or not poverty is a cause of land degradation (Gerber et al. 2014; Vu et al. 2014; Way 2016) (Section 4.7.1). The literature measuring the extent to which desertification contributed to poverty globally is lacking: the related literature remains qualitative or correlational (Barbier and Hochard 2016). At the local level, on the other hand, there is limited evidence and high agreement that desertification increased multidimensional poverty. For example, Diao and Sarpong (2011) estimated that land degradation lowered agricultural incomes in Ghana by 4.2 billion USD between 2006 and 2015, increasing the national poverty rate by 5.4% in 2015. Land degradation increased the probability of households becoming poor by 35% in Malawi and 48% in Tanzania (Kirui 2016). Desertification in China was found to have resulted in substantial losses in income, food production and jobs (Jiang et al. 2014). On the other hand, Ge et al. (2015) indicated that desertification was positively associated with growing incomes in Inner Mongolia in China in the short run since no costs were incurred for SLM, while in the long run higher incomes allowed allocation of more investments to reduce desertification. This relationship corresponds to the Environmental Kuznets Curve, which posits that environmental degradation initially rises and subsequently falls with rising income (e.g., Stern 2017). There is limited evidence on the validity of this hypothesis regarding desertification.

The frequency and intensity of dust storms are increasing due to land-use and land-cover changes and climate-related factors (Section 2.4) particularly in some regions of the world such as the Arabian Peninsula (Jish Prakash et al. 2015; Yu et al. 2015; Gherboudj et al. 2017; Notaro et al. 2013; Yu et al. 2013; Alobaidi et al. 2017; Maghrabi et al. 2011; Almazroui et al. 2018) and broader Middle East (Rashki et al. 2012; Türkeş 2017; Namdari et al. 2018) as well as Central Asia (Indoitu et al. 2015; Xi and Sokolik 2015), with growing negative impacts on human health (high confidence) (Díaz et al. 2017; Goudarzi et al. 2017; Goudie 2014; Samoli et al. 2011). Dust storms transport particulate matter, pollutants, pathogens and potential allergens that are dangerous for human health over long distances (Goudie and Middleton 2006; Sprigg 2016). Particulate matter (PM; that is, the suspended particles in the air of up to 10 micrometres (PM10) or less in size), have damaging effects on human health (Díaz et al. 2017; Goudarzi et al. 2017; Goudie 2014; Samoli et al. 2011). The health effects of dust storms are largest in areas in the immediate vicinity of their origin, primarily the Sahara Desert, followed by Central and eastern Asia, the Middle East and Australia (Zhang et al. 2016), however, there is robust evidence showing that the negative health effects of dust storms reach a much wider area (Bennett et al. 2006; Díaz et al. 2017; Kashima et al. 2016; Lee et al. 2014; Samoli et al. 2011; Zhang et al. 2016). The primary health effects of dust storms include damage to the respiratory and cardiovascular systems (Goudie 2013). Dust particles with a diameter smaller than 2.5μm were associated with global cardiopulmonary mortality of about 402,000 people in 2005, with 3.47 million years of life lost in that single year (Giannadaki et al. 2014). Although globally only 1.8% of cardiopulmonary deaths were caused by dust storms, in the countries of the Sahara region, Middle East, South and East Asia, dust storms were suggested to be the cause of 15–50% of all cardiopulmonary deaths (Giannadaki et al. 2014). A 10 μgm-³ increase in PM10 dust particles was associated with mean increases in non-accidental mortality from 0.33% to 0.51% across different calendar seasons in China, Japan and South Korea (Kim et al. 2017). The percentage of all-cause deaths attributed to fine particulate matter in Iranian cities affected by Middle Eastern dust storms (MED) was 0.56–5.02%, while the same percentage for non-affected cities was 0.16–4.13% (Hopke et al. 2018). Epidemics of meningococcal meningitis occur in the Sahelian region during the dry seasons with dusty conditions (Agier et al. 2012; Molesworth et al. 2003). Despite a strong concentration of dust storms in the Sahel, North Africa, the Middle East and Central Asia, there is relatively little research on human health impacts of dust storms in these regions. More research on health impacts and related costs of dust storms, as well as on public health response measures, can help in mitigating these health impacts.

Environmental issues such as desertification and impacts of climate change have been increasingly investigated through a gender lens (Bose (n.d.); Broeckhoven and Cliquet 2015; Kaijser and Kronsell 2014; Kiptot et al. 2014; Villamor and van Noordwijk 2016). There is medium evidence and high agreement that women will be impacted more than men by environmental degradation (Arora-Jonsson 2011; Gurung et al. 2006) (Cross-Chapter Box 11 in Chapter 7). Socially structured gender-specific roles and responsibilities, daily activities, access and control over resources, decision-making and opportunities lead men and women to interact differently with natural resources and landscapes. For example, water scarcity affected women more than men in rural Ghana as they had to spend more time in fetching water, which has implications on time allocations for other activities (Ahmed et al. 2016). Despite the evidence pointing to differentiated impact of environmental degradation on women and men, gender issues have been marginally addressed in many land restoration and rehabilitation efforts, which often remain gender-blind. Although there is robust evidence on the location-specific impacts of climate change and desertification on gender equality, there is limited evidence on the gender-related impacts of land restoration and rehabilitation activities. Women are usually excluded from local decision-making on actions regarding desertification and climate change. Socially constructed gender-specific roles and responsibilities are not static because they are shaped by other factors such as wealth, age, ethnicity and formal education (Kaijser and Kronsell 2014; Villamor et al. 2014). Hence, women’s and men’s environmental knowledge and priorities for restoration often differ (Sijapati Basnett et al. 2017). In some areas where sustainable land options (e.g., agroforestry) are being promoted, women were not able to participate due to culturally embedded asymmetries in power relations between men and women (Catacutan and Villamor 2016). Nonetheless women, particularly in the rural areas, remain heavily involved in securing food for their households. Food security for them is associated with land productivity and women’s contribution to address desertification is crucial.

Reduced water retention capacity of degraded soils amplifies floods (de la Paix et al. 2011), reinforces degradation processes through soil erosion, and reduces annual intake of water to aquifers, exacerbating existing water scarcities (Le Roux et al. 2017; Cano et al. 2018). Reduced vegetation cover and more intense dust storms were found to intensify droughts (Cook et al. 2009). Moreover, secondary salinisation in the irrigated drylands often requires leaching with considerable amounts of water (Greene et al. 2016; Wichelns and Qadir 2015). Thus, different types of soil degradation increase water scarcity both through lower water quantity and quality (Liu et al. 2017; Liu et al. 2016c). All these processes reduce water availability for other needs. In this context, climate change will further intensify water scarcity in some dryland areas and increase the frequency of droughts (medium confidence) (IPCC 2013; Zheng et al. 2018) (Section 2.2). Higher water scarcity may imply growing use of wastewater effluents for irrigation (Pedrero et al. 2010). The use of untreated wastewater exacerbates soil degradation processes (Tal 2016; Singh et al. 2004; Qishlaqi et al. 2008; Hanjra et al. 2012), in addition to negative human health impacts (Faour-Klingbeil and Todd 2018; Hanjra et al. 2012). Climate change will thus amplify the need for integrated land and water management for sustainable development.

Desertification leads to conditions that favour the production of dust storms (high confidence) (Section 3.3.1). There is robust evidence and high agreement that dust storms negatively affect the operational potential of solar and wind power harvesting equipment through dust deposition, reduced reach of solar radiation and increasing blade-surface roughness, and can also reduce effective electricity distribution in high-voltage transmission lines (Zidane et al. 2016; Costa et al. 2016; Lopez-Garcia et al. 2016; Maliszewski et al. 2012; Mani and Pillai 2010; Mejia and Kleissl 2013; Mejia et al. 2014; Middleton 2017; Sarver et al. 2013; Kaufman et al. 2002; Kok et al. 2018). Direct exposure to desert dust storm can reduce energy generation efficiency of solar panels by 70–80% in one hour (Ghazi et al. 2014). (Saidan et al. 2016) indicated that in the conditions of Baghdad, Iraq, one month’s exposure to weather reduced the efficiency of solar modules by 18.74% due to dust deposition. In the Atacama desert, Chile, one month’s exposure reduced thin-film solar module performance by 3.7–4.8% (Fuentealba et al. 2015). This has important implications for climate change mitigation efforts using the expansion of solar and wind energy generation in dryland areas for substituting fossil fuels. Abundant access to solar energy in many dryland areas makes them high-potential locations for the installation of solar energy generating infrastructure. Increasing desertification, resulting in higher frequency and intensity of dust storms imposes additional costs for climate change mitigation through deployment of solar and wind energy harvesting facilities in dryland areas. Most frequently used solutions to this problem involve physically wiping or washing the surface of solar devices with water. These result in additional costs and excessive use of already scarce water resources and labour (Middleton 2017). The use of special coatings on the surface of solar panels can help prevent the deposition of dusts (Costa et al. 2016; Costa et al. 2018; Gholami et al. 2017).

Dust storms and movement of sand dunes often threaten the safety and operation of railway and road infrastructure in arid and hyper-arid areas, and can lead to road and airport closures due to reductions in visibility. For example, the dust storm on 10 March 2009 over Riyadh was assessed to be the strongest in the previous two decades in Saudi Arabia, causing limited visibility, airport shutdown and damages to infrastructure and environment across the city (Maghrabi et al. 2011). There are numerous historical examples of how moving sand dunes led to the forced decommissioning of early railway lines built in Sudan, Algeria, Namibia and Saudi Arabia in the late 19th and early 20th century (Bruno et al. 2018). Currently, the highest concentrations of railways vulnerable to sand movements are located in north-western China, Middle East and North Africa (Bruno et al. 2018; Cheng and Xue 2014). In China, sand dune movements are periodically disrupting the railway transport on the Linhai–Ceke line in north-western China and on the Lanzhou–Xinjiang High-speed Railway in western China, with considerable clean-up and maintenance costs (Bruno et al. 2018; Zhang et al. 2010). There are large-scale plans for expansion of railway networks in arid areas of China, Central Asia, North Africa, the Middle East, and eastern Africa. For example, The Belt and Road Initiative promoted by China, the Gulf Railway project by the Cooperation Council for the Arab States of the Gulf or Lamu Port–South Sudan–Ethiopia Transport (LAPSSET) Corridor in Eastern Africa. These investments have long-term return and operation periods. Their construction and associated engineering solutions will therefore benefit from careful consideration of potential desertification and climate change effects on sand storms and dune movements.

There is low confidence in climate change and desertification leading to violent conflicts. There is medium evidence and low agreement that climate change and desertification contribute to already existing conflict potentials (Herrero 2006; von Uexkull et al. 2016; Theisen 2017; Olsson 2017; Wischnath and Buhaug 2014) (Section 4.7.3). To illustrate, Hsiang et al. (2013) found that each one standard deviation increase in temperature or rainfall was found to increase interpersonal violence by 4% and intergroup conflict by 14% (Hsiang et al. 2013). However, this conclusion was disputed by Buhaug et al. (2014), who found no evidence linking climate variability to violent conflict after replicating Hsiang et al. (2013) by studying only violent conflicts. Almer et al. (2017) found that a one standard deviation increase in dryness raised the likelihood of riots in Sub-Saharan African countries by 8.3% during the 1990–2011 period. On the other hand, Owain and Maslin (2018) found that droughts and heatwaves were not significantly affecting the level of regional conflict in East Africa. Similarly, it was suggested that droughts and desertification in the Sahel played a relatively minor role in the conflicts in the Sahel in the 1980s, with the major reasons for the conflicts during this period being political, especially the marginalisation of pastoralists (Benjaminsen 2016), corruption and rent-seeking (Benjaminsen et al. 2012). Moreover, the role of environmental factors as the key drivers of conflicts was questioned in the case of Sudan (Verhoeven 2011) and Syria (De Châtel 2014). Selection bias, when the literature focuses on the same few regions where conflicts occurred and relates them to climate change, is a major shortcoming, as it ignores other cases where conflicts did not occur (Adams et al. 2018) despite degradation of the natural resource base and extreme weather events.

Environmentally induced migration is complex and accounts for multiple drivers of mobility as well as other adaptation measures undertaken by populations exposed to environmental risk (high confidence). There is medium evidence and low agreement that climate change impacts migration. The World Bank (2018) predicted that 143 million people would be forced to move internally by 2050 if no climate action is taken. Focusing on asylum seekers alone, rather than the total number of migrants, Missirian and Schlenker (2017) predict that asylum applications to the European Union will increase from 28% (98,000 additional asylum applications per year) up to 188% (660,000 additional applications per year) depending on the climate scenario by 2100. While the modelling efforts have greatly improved over the years (Hunter et al. 2015; McLeman 2011; Sherbinin and Bai 2018) and in particular, these recent estimates provide an important insight into potential future developments, the quantitative projections are still based on the number of people exposed to risk rather than the number of people who would actually engage in migration as a response to this risk (Gemenne 2011; McLeman 2013) and they do not take into account individual agency in migration decision nor adaptive capacities of individuals (Hartmann 2010; Kniveton et al. 2011; Piguet 2010) (see Section 3.6.2 discussing migration as a response to desertification). Accordingly, the available micro-level evidence suggests that climate-related shocks are one of the many drivers of migration (Adger et al. 2014; London Government Office for Science and Foresight 2011; Melde et al. 2017), but the individual responses to climate risk are more complex than commonly assumed (Gray and Mueller 2012a). For example, despite strong focus on natural disasters, neither flooding (Gray and Mueller 2012b; Mueller et al. 2014) nor earthquakes (Halliday 2006) were found to induce long-term migration; but instead, slow-onset changes, especially those provoking crop failures and heat stress, could affect household or individual migration decisions (Gray and Mueller 2012a; Missirian and Schlenker 2017; Mueller et al. 2014). Out-migration from drought-prone areas has received particular attention (de Sherbinin et al. 2012; Ezra and Kiros 2001). A substantial body of literature suggests that households engage in local or internal migration as a response to drought (Findlay 2011; Gray and Mueller 2012a), while international migration decreases with drought in some contexts (Henry et al. 2004), but might increase in contexts where migration networks are well established (Feng et al. 2010; Nawrotzki and DeWaard 2016; Nawrotzki et al. 2015, 2016). Similarly, the evidence is not conclusive with respect to the effect of environmental drivers, in particular desertification, on mobility. While it has not consistently entailed out-migration in the case of Ecuadorian Andes (Gray 2009, 2010), environmental and land degradation increased mobility in Kenya and Nepal (Gray 2011; Massey et al. 2010), but marginally decreased mobility in Uganda (Gray 2011). These results suggest that in some contexts, environmental shocks actually undermine households’ financial capacity to undertake migration (Nawrotzki and Bakhtsiyarava 2017), especially in the case of the poorest households (Barbier and Hochard 2018; Koubi et al. 2016; Kubik and Maurel 2016; McKenzie and Yang 2015). Adding to the complexity, migration, especially to frontier areas, by increasing pressure on land and natural resources, might itself contribute to environmental degradation at the destination (Hugo 2008; IPBES 2018a; McLeman 2017). The consequences of migration can also be salient in the case of migration to urban or peri-urban areas; indeed, environmentally induced migration can add to urbanisation (Section 3.6.2.2), often exacerbating problems related to poor infrastructure and unemployment.

Pastoral production systems occupy a significant portion of the world (Rass 2006; Dong 2016). Food insecurity among pastoral households is often high (Gomes 2006) (Section 3.1.3). The Sahelian droughts of the 1970s–1980s provided an example of how droughts could affect livestock resources and crop productivity, contributing to hunger, out-migration and suffering for millions of pastoralists (Hein and De Ridder 2006; Molua and Lambi 2007). During these Sahelian droughts low and erratic rainfall exacerbated desertification processes, leading to ecological changes that forced people to use marginal lands and ecosystems. Similarly, the rate of rangeland degradation is now increasing because of environmental changes and overexploitation of resources (Kassahun et al. 2008; Vetter 2005). Desertification coupled with climate change is negatively affecting livestock feed and grazing species (Hopkins and Del Prado 2007), changing the composition in favour of species with low forage quality, ultimately reducing livestock productivity (D’Odorico et al. 2013; Dibari et al. 2016) and increasing livestock disease prevalence (Thornton et al. 2009). There is robust evidence and high agreement that weak adaptive capacity, coupled with negative effects from other climate-related factors, are predisposing pastoralists to increased poverty from desertification and climate change globally (López-i-Gelats et al. 2016; Giannini et al. 2008; IPCC 2007). On the other hand, misguided policies such as enforced sedentarisation, and in certain cases protected area delineation (fencing), which restrict livestock mobility have hampered optimal use of grazing land resources (Du 2012). Such policies have led to degradation of resources and out-migration of people in search of better livelihoods (Gebeye 2016; Liao et al. 2015). Restrictions on the mobile lifestyle are reducing the resilient adaptive capacity of pastoralists to natural hazards including extreme and variable weather conditions, drought and climate change (Schilling et al. 2014). Furthermore, the exacerbation of the desertification phenomenon due to agricultural intensification (D’Odorico et al. 2013) and land fragmentation caused by encroachment of agriculture into rangelands (Otuoma et al. 2009; Behnke and Kerven 2013) is threatening pastoral livelihoods. For example, commercial cotton (Gossypium hirsutum) production is crowding out pastoral systems in Benin (Tamou et al. 2018). Food shortages and the urgency to produce enough crop for public consumption are leading to the encroachment of agriculture into productive rangelands and those converted rangelands are frequently prime lands used by pastoralists to produce feed and graze their livestock during dry years (Dodd 1994). The sustainability of pastoral systems is therefore coming into question because of social and political marginalisation of those systems (Davies et al. 2016) and also because of the fierce competition they are facing from other livelihood sources such as crop farming (Haan et al. 2016).

Following the conceptual framework developed in the Special Report on extreme events (SREX) (IPCC 2012), future risks are assessed by examining changes in exposure (that is, presence of people; livelihoods; species or ecosystems; environmental functions, service, and resources; infrastructure; or economic, social or cultural assets; see Glossary), changes in vulnerability (that is, propensity or predisposition to be adversely affected; see Glossary) and changes in the nature and magnitude of hazards (that is, potential occurrence of a natural or human-induced physical event that causes damage; see Glossary). Climate change is expected to further exacerbate the vulnerability of dryland ecosystems to desertification by increasing PET globally (Sherwood and Fu 2014). Temperature increases between 2°C and 4°C are projected in drylands by the end of the 21st century under RCP4.5 and RCP8.5 scenarios, respectively (IPCC 2013). An assessment by Carrão et al. 2017 showed an increase in drought hazards by late-century (2071–2099) compared to a baseline (1971–2000) under high RCPs in drylands around the Mediterranean, south-eastern Africa, and southern Australia. In Latin America, Morales et al. (2011) indicated that areas affected by drought will increase significantly by 2100 under SRES scenarios A2 and B2. The countries expected to be affected include Guatemala, El Salvador, Honduras and Nicaragua. In CMIP5 scenarios, Mediterranean types of climate are projected to become drier (Alessandri et al. 2014; Polade et al. 2017), with the equatorward margins being potentially replaced by arid climate types (Alessandri et al. 2014). Globally, climate change is predicted to intensify the occurrence and severity of droughts (medium confidence) (Dai 2013; Sheffield and Wood 2008; Swann et al. 2016; Wang 2005; Zhao and Dai 2015; Carrão et al. 2017; Naumann et al. 2018) (Section 2.2). Ukkola et al. (2018) showed large discrepancies between CMIP5 models for all types of droughts, limiting the confidence that can be assigned to projections of drought.

Rangeland management systems such as sustainable grazing approaches and re-vegetation increase rangeland productivity (high confidence) (Table 6.5). Open grassland, savannah and woodland are home to the majority of world’s livestock production (Safriel et al. 2005). Within these drylands areas, prevailing grazing and fire regimes play an important role in shaping the relative abundance of trees versus grasses (Scholes and Archer 1997; Staver et al. 2011; Stevens et al. 2017), as well as the health of the grass layer in terms of primary production, species richness and basal cover (the propotion of the plant that is in the soil) (Plaza-Bonilla et al. 2015; Short et al. 2003). This in turn influences levels of soil erosion, soil nutrients, secondary production and additional ecosystem services (Divinsky et al. 2017; Pellegrini et al. 2017). A further set of drivers, including soil type, annual rainfall and changes in atmospheric CO2 may also define observed rangeland structure and composition (Devine et al. 2017; Donohue et al. 2013), but the two principal factors that pastoralists can manage are grazing and fire, by altering their frequency, type and intensity.

The impact of grazing and fire regimes on biodiversity, soil nutrients, primary production and further ecosystem services is not constant and varies between locations (Divinsky et al. 2017; Fleischner 1994; van Oijen et al. 2018). Trade-offs may therefore need to be considered to ensure that rangeland diversity and production are resilient to climate change (Plaza-Bonilla et al. 2015; van Oijen et al. 2018). In certain locations, even light to moderate grazing has led to a significant decrease in the occurrence of particular species, especially forbs (O’Connor et al. 2011; Scott-shaw and Morris 2015). In other locations, species richness is only significantly impacted by heavy grazing and is able to withstand light to moderate grazing (Divinsky et al. 2017). A context specific evaluation of how grazing and fire impact particular species may therefore be required to ensure the persistence of target species over time (Marty 2005). A similar trade-off may need to be considered between soil carbon sequestration and livestock production. As noted by Plaza-Bonilla et al. (2015) increasing grazing pressure has been found to increase SOC stocks in some locations, and decrease them in others. Where it has led to a decrease in soil carbon stocks, for example in Mongolia (Han et al. 2008) and Ethiopia (Bikila et al. 2016), trade-offs between carbon sequestration and the value of livestock to local livelihoods need be considered.

The encroachment of open grassland and savannah ecosystems by woody species has occurred for at least the past 100 years (Archer et al. 2017; O’Connor et al. 2014; Schooley et al. 2018). Dependent on the type and intensity of encroachment, it may lead to a net loss of ecosystem services and be viewed as a form of desertification (Dougill et al. 2016; O’Connor et al. 2014). However, there are circumstances where bush encroachment may lead to a net increase in ecosystem services, especially at intermediate levels of encroachment, where the ability of the landscape to produce fodder for livestock is retained, while the production of wood and associated products increases (Eldridge et al. 2011; Eldridge and Soliveres 2014). This may be particularly important in regions such as southern Africa and India where over 65% of rural households depend on fuelwood from surrounding landscapes as well as livestock production (Komala and Prasad 2016; Makonese et al. 2017; Shackleton and Shackleton 2004).

This variable relationship between the level of encroachment, carbon stocks, biodiversity, provision of water and pastoral value (Eldridge and Soliveres 2014) can present a conundrum to policymakers, especially when considering the goals of three Rio Conventions: UNFCCC, UNCCD and UNCBD. Clearing intense bush encroachment may improve species diversity, rangeland productivity, the provision of water and decrease desertification, thereby contributing to the goals of the UNCBD and UNCCD as well as the adaptation aims of the UNFCCC. However, it would lead to the release of biomass carbon stocks into the atmosphere and potentially conflict with the mitigation aims of the UNFCCC.

For example, Smit et al. (2015) observed an average increase in above-ground woody carbon stocks of 44 tC ha–1 in savannahs in northern Namibia. However, since bush encroachment significantly inhibited livestock production, there are often substantial efforts to clear woody species (Stafford-Smith et al. 2017). Namibia has a national programme, currently in its early stages, aimed at clearing woody species through mechanical measures (harvesting of trees) as well as the application of arboricides (Smit et al. 2015). However, the long-term success of clearance and subsequent improved fire and grazing management remains to be evaluated, especially restoration back towards an ‘original open grassland state’. For example, in northern Namibia, the rapid reestablishment of woody seedlings has raised questions about whether full clearance and restoration is possible (Smit et al. 2015). In arid landscapes, the potential impact of elevated atmospheric CO2 (Donohue et al. 2013; Kgope et al. 2010) and opportunity to implement high-intensity fires that remove woody species and maintain rangelands in an open state has been questioned (Bond and Midgley 2000). If these drivers of woody plant encroachment cannot be addressed, a new form of ‘emerging ecosystem’ (Milton 2003) may need to be explored that includes both improved livestock and fire management as well as the utilisation of biomass as a long-term commodity and source of revenue (Smit et al. 2015). Initial studies in Namibia and South Africa (Stafford-Smith et al. 2017) indicate that there may be good opportunity to produce sawn timber, fencing poles, fuelwood and commercial energy, but factors such as the cost of transport can substantially influence the financial feasibility of implementation.

Dust and sand storms have a considerable impact on natural and human systems (Sections 3.4.1 and 3.4.2). Application of sand dune stabilisation techniques contributes to reducing sand and dust storms (high confidence). Using a number of methods, sand dune stabilisation aims to avoid and reduce the occurrence of dust and sand storms (Mainguet and Dumay 2011). Mechanical techniques include building palisades to prevent the movement of sand and reduce sand deposits on infrastructure. Chemical methods include the use of calcium bentonite or using silica gel to fix mobile sand (Aboushook et al. 2012; Rammal and Jubair 2015). Biological methods include the use of mulch to stabilise surfaces (Sebaa et al. 2015; Yu et al. 2004) and establishing permanent plant cover using pasture species that improve grazing at the same time (Abdelkebir and Ferchichi 2015; Zhang et al. 2015) (Section 3.7.1.3). When the dune is stabilised, woody perennials are introduced that are selected according to climatic and ecological conditions (FAO 2011). For example, such re-vegetation processes have been implemented on the shifting dunes of the Tengger Desert in northern China leading to the stabilisation of sand and the sequestration of up to 10 tC ha–1 over a period of 55 years (Yang et al. 2014).

Soil salinity and sodicity can severely limit the growth and productivity of crops (Jan et al. 2017) and lead to a decrease in available arable land. Leaching and drainage provides a possible solution, but can be prohibitively expensive. An alternative, more economical option, is the growth of halophytes (plants that are adapted to grow under highly saline conditions) that allow saline land to be used in a productive manner (Qadir et al. 2000). The biomass produced can be used as forage, food, feed, essential oils, biofuel, timber, or fuelwood (Chughtai et al. 2015; Mahmood et al. 2016; Sharma et al. 2016). A further co-benefit is the opportunity to mitigate climate change through the enhancement of terrestrial carbon stocks as land is re-vegetated (Dagar et al. 2014; Wicke et al. 2013). The combined use of salt-tolerant crops, improved irrigation practices, chemical remediation measures and appropriate mulch and compost is effective in reducing the impact of secondary salinisation (medium confidence).

In Pakistan, where about 6.2 Mha of agricultural land is affected by salinity, pioneering work on utilising salt-tolerant plants for the re-vegetation of saline lands (biosaline agriculture) was done in the early 1970s (NIAB 1997). A number of local and exotic varieties were initially screened for salt tolerance in lab – and greenhouse-based studies, and then distributed to similar saline areas (Ashraf et al. 2010). These included tree species (Acacia ampliceps, Acacia nilotica, Eucalyptus camaldulensis, Prosopis juliflora, Azadirachta indica) (Awan and Mahmood 2017), forage plants (Leptochloa fusca, Sporobolus arabicus, Brachiaria mutica, Echinochloa sp., Sesbania and Atriplex spp.) and crop species including varieties of barley (Hordeum vulgare), cotton, wheat (Triticum aestivum) and Brassica spp. (Mahmood et al. 2016) as well as fruit crops in the form of date palm (Phoenix dactylifera) that has high salt tolerance with no visible adverse effects on seedlings (Yaish and Kumar 2015; Al-Mulla et al. 2013; Alrasbi et al. 2010). Pomegranate (Punica granatum L.) is another fruit crop of moderate to high salt tolerance. Through regulating growth form and nutrient balancing, it can maintain water content, chlorophyll fluorescence and enzyme activity at normal levels (Ibrahim 2016; Okhovatian-Ardakani et al. 2010).

In India and elsewhere, tree species including Prosopis juliflora, Dalbergia sissoo, and Eucalyptus tereticornis have been used to re-vegetate saline land. Certain biofuel crops in the form of Ricinus communis (Abideen et al. 2014), Euphorbia antisyphilitica (Dagar et al. 2014), Karelinia caspia (Akinshina et al. 2016) and Salicornia spp. (Sanandiya and Siddhanta 2014) are grown in saline areas, and Panicum turgidum (Koyro et al. 2013) and Leptochloa fusca (Akhter et al. 2003) have been grown as fodder crop on degraded soils with brackish water. In China, intense efforts are being made on the use of halophytes (Sakai et al. 2012; Wang et al. 2018). These examples reveal that there is great scope for saline areas to be used in a productive manner through the utilisation of halophytes. The most productive species often have yields equivalent to conventional crops, at salinity levels matching even that of seawater.

Desertification limits the choice of potential climate change mitigation and adaptation response options by reducing climate change adaptive capacities. Furthermore, many additional factors, for example, a lack of access to markets or insecurity of land tenure, hinder the adoption of SLM. These factors are largely beyond the control of individuals or local communities and require broader policy interventions (Section 3.6.3). Nevertheless, local collective action and ILK are still crucial to the ability of households to respond to the combined challenge of climate change and desertification. Raising awareness, capacity building and development to promote collective action and indigenous and local knowledge contribute to avoiding, reducing and reversing desertification under changing climate.

Livelihood diversification through non-farm employment increases the resilience of rural households against desertification and extreme weather events by diversifying their income and consumption (high confidence). Moreover, it can provide the funds to invest into SLM (Belay et al. 2017; Bryan et al. 2009; Dumenu and Obeng 2016; Salik et al. 2017; Shiferaw et al. 2009). Access to non-agricultural employment is especially important for poorer pastoral households as their small herd sizes make them less resilient to drought (Fratkin 2013; Lybbert et al. 2004). However, access to alternative opportunities is limited in the rural areas of many developing countries, especially for women and marginalised groups who lack education and social networks (Reardon et al. 2008).

Migration is frequently used as an adaptation strategy to environmental change (medium confidence). Migration is a form of livelihood diversification and a potential response option to desertification and increasing risk to agricultural livelihoods under climate change (Walther et al. 2002). Migration can be short-term (e.g., seasonal) or long-term, internal within a country or international. There is medium evidence showing rural households responding to desertification and droughts through all forms of migration, for example: during the Dust Bowl in the USA in the 1930s (Hornbeck 2012); during droughts in Burkina Faso in the 2000s (Barbier et al. 2009); in Mexico in the 1990s (Nawrotzki et al. 2016); and by the Aymara people of the semi-arid Tarapacá region in Chile between 1820 and 1970, responding to declines in rainfall and growing demands for labour outside the region (Lima et al. 2016). There is robust evidence and high agreement showing that migration decisions are influenced by a complex set of different factors, with desertification and climate change playing relatively lesser roles (Liehr et al. 2016) (Section 3.4.2). Barrios et al. (2006) found that urbanisation in Sub-Saharan Africa was partially influenced by climatic factors during the 1950–2000 period, in parallel to liberalisation of internal restrictions on labour movements: each 1% reduction in rainfall was associated with a 0.45% increase in urbanisation. This migration favoured more industrially diverse urban areas in Sub-Saharan Africa (Henderson et al. 2017), because they offer more diverse employment opportunities and higher wages. Similar trends were also observed in Iran in response to water scarcity (Madani et al. 2016).

However, migration involves some initial investments. For this reason, reductions in agricultural incomes due to climate change or desertification have the potential to decrease out-migration among the poorest agricultural households, who become less able to afford migration (Cattaneo and Peri 2016), thus increasing social inequalities. There is medium evidence and high agreement that households with migrant worker members are more resilient against extreme weather events and environmental degradation compared to non-migrant households, who are more dependent on agricultural income (Liehr et al. 2016; Salik et al. 2017; Sikder and Higgins 2017). Remittances from migrant household members potentially contribute to SLM adoptions, however, substantial out-migration was also found to constrain the implementation of labour-intensive land management practices (Chen et al. 2014; Liu et al. 2016a).

Policy responses to combat desertification take numerous forms (Marques et al. 2016). Below we discuss major policy responses consistently highlighted in the literature in connection with SLM and climate change, because these response options were found to strengthen adaptation capacities and to contribute to climate change mitigation. They include improving market access, empowering women, expanding access to agricultural advisory services, strengthening land tenure security, payments for ecosystem services, decentralised natural resource management, investing into research and monitoring of desertification and dust storms, and investing into modern renewable energy sources.

Policies aiming at improving market access, that is the ability to access output and input markets at lower costs, help farmers and livestock producers earn more profit from their produce. Increased profits both motivate and enable them to invest more in SLM. Higher access to input, output and credit markets was consistently found as a major factor in the adoption of SLM practices in a wide number of settings across the drylands (medium confidence) (Aw-Hassan et al. 2016; Gebreselassie et al. 2016; Mythili and Goedecke 2016; Nkonya and Anderson 2015; Sow et al. 2016). Lack of access to credit limits adjustments and agricultural responses to the impacts of desertification under changing climate, with long-term consequences for the livelihoods and incomes, as was shown during the North American Dust Bowl of the 1930s (Hornbeck 2012). Government policies aimed at improving market access usually involve constructing and upgrading rural–urban transportation infrastructure and agricultural value chains, such as investments into construction of local markets, abattoirs and cold storage warehouses, as well as post-harvest processing facilities (McPeak et al. 2006). However, besides infrastructural constraints, providing improved access often involves relieving institutional constraints to market access (Little 2010), such as improved coordination of cross-border food safety and veterinary regulations (Ait Hou et al. 2015; Keiichiro et al. 2015; McPeak et al. 2006; Unnevehr 2015), and availability and access to market information systems (Bobojonov et al. 2016; Christy et al. 2014; Nakasone et al. 2014).

Women’s empowerment. A greater emphasis on understanding gender-specific differences over land use and land management practices as an entry point can make land restoration projects more successful (medium confidence) (Broeckhoven and Cliquet 2015; Carr and Thompson 2014; Catacutan and Villamor 2016; Dah-gbeto and Villamor 2016). In relation to representation and authority to make decisions in land management and governance, women’s participation remains lacking particularly in the dryland regions. Thus, ensuring women’s rights means accepting women as equal members of the community and citizens of the state (Nelson et al. 2015). This includes equitable access of women to resources (including extension services), networks, and markets. In areas where socio-cultural norms and practices devalue women and undermine their participation, actions for empowering women will require changes in customary norms, recognition of women’s (land) rights in government policies, and programmes to assure that their interests are better represented (Section 1.4.2 and Cross-Chapter Box 11 in Chapter 7). In addition, several novel concepts are recently applied for an in-depth understanding of gender in relation to science–policy interface. Among these are the concepts of intersectionality, that is, how social dimensions of identity and gender are bound up in systems of power and social institutions (Thompson-Hall et al. 2016), bounded rationality for gendered decision-making, related to incomplete information interacting with limits to human cognition leading to judgement errors or objectively poor decision making (Villamor and van Noordwijk 2016), anticipatory learning for preparing for possible contingencies and consideration of long-term alternatives (Dah-gbeto and Villamor 2016) and systematic leverage points for interventions that produce, mark, and entrench gender inequality within communities (Manlosa et al. 2018), which all aim to improve gender equality within agroecological landscapes through a systems approach.

Education and expanding access to agricultural services. Providing access to information about SLM practices facilitates their adoption (medium confidence) (Kassie et al. 2015; Nkonya et al. 2015; Nyanga et al. 2016). Moreover, improving the knowledge of climate change, capacity building and development in rural areas can help strengthen climate change adaptive capacities (Berman et al. 2012; Chen et al. 2018; Descheemaeker et al. 2018; Popp et al. 2009; Tambo 2016; Yaro et al. 2015). Agricultural initiatives to improve the adaptive capacities of vulnerable populations were more successful when they were conducted through reorganised social institutions and improved communication, for example, in Mozambique (Osbahr et al. 2008). Improved communication and education could be facilitated by wider use of new information and communication technologies (ICTs) (Peters et al. 2015). Investments into education were associated with higher adoption of soil conservation measures, for example, in Tanzania (Tenge et al. 2004). Bryan et al. (2009) found that access to information was the prominent facilitator of climate change adaptation in Ethiopia. However, resource constraints of agricultural services, and disconnects between agricultural policy and climate policy can hinder the dissemination of climate-smart agricultural technologies (Morton 2017). Lack of knowledge was also found to be a significant barrier to implementation of soil rehabilitation programmes in the Mediterranean region (Reichardt 2010). Agricultural services will be able to facilitate SLM best when they also serve as platforms for sharing indigenous and local knowledge and farmer innovations (Mapfumo et al. 2016). Participatory research initiatives conducted jointly with farmers have higher chances of resulting in technology adoption (Bonney et al. 2016; Rusike et al. 2006; Vente et al. 2016). Moreover, rural advisory services are often more successful in disseminating technological innovations when they adopt commodity/value chain approaches, remain open to engagement in input supply, make use of new opportunities presented by ICTs, facilitate mutual learning between multiple stakeholders (Morton 2017), and organise science and SLM information in a location-specific manner for use in education and extension (Bestelmeyer et al. 2017).

Despite policy responses for combating desertification, other factors will put strong pressures on the land, including climate change and growing food demands, as well as the need to reduce poverty and strengthen food security (Cherlet et al. 2018) (Sections 6.1.4 and 7.2.2). Sustainable development of drylands and their resilience to combined challenges of desertification and climate change will thus also depend on the ability of governments to promote policies for economic diversification within agriculture and in non-agricultural sectors in order make dryland areas less vulnerable to desertification and climate change.

Investing into irrigation. Investments into expanding irrigation in dryland areas can help increase the resilience of agricultural production to climate change, improve labour productivity and boost production and income revenue from agriculture and livestock sectors (Geerts and Raes 2009; Olayide et al. 2016; Oweis and Hachum 2006). This is particularly true for Sub-Saharan Africa, where currently only 6% of the cultivated areas are irrigated (Nkonya et al. 2016b). While renewable groundwater resources could help increase the share of irrigated land to 20.5–48.6% of croplands in the region (Altchenko and Villholth 2015). On the other hand, over-extraction of groundwaters, mainly for irrigating crops, is becoming an important environmental problem in many dryland areas (Cherlet et al. 2018), requiring careful design and planning of irrigation expansion schemes and use of water-efficient irrigation methods (Bjornlund et al. 2017; Woodhouse et al. 2017). For example, in Saudi Arabia, improving the efficiency of water management, for example through the development of aquifers, water recycling and rainwater harvesting, is part of a suite of policy actions to combat desertification (Bazza, et al. 2018; Kingdom of Saudi Arabia 2016). The expansion of irrigation to riverine areas, crucial for dry season grazing of livestock, needs to consider the income from pastoral activities, which is not always lower than income from irrigated crop production (Behnke and Kerven 2013). Irrigation development could be combined with the deployment of clean-energy technologies in economically viable ways (Chandel et al. 2015). For example, solar-powered drip irrigation was found to increase household agricultural incomes in Benin (Burney et al. 2010). The sustainability of irrigation schemes based on solar-powered extraction of groundwaters depends on measures to avoid over-abstraction of groundwater resources and associated negative environmental impacts (Closas and Rap 2017).

Soil erosion is a major form of desertification occurring in varying degrees in all dryland areas across the world (Section 3.2), with negative effects on dryland ecosystems (Section 3.4). Climate change is projected to increase soil erosion potential in some dryland areas through more frequent heavy rainfall events and rainfall variability (see Section 3.5.2 for a more detailed assessment) (Achite and Ouillon 2007; Megnounif and Ghenim 2016; Vachtman et al. 2013; Zhang and Nearing 2005). There are numerous soil conservation measures that can help reduce soil erosion (Section 3.6.1). Such soil management measures include afforestation and reforestation activities, rehabilitation of degraded forests, erosion control measures, prevention of overgrazing, diversification of crop rotations, and improvement in irrigation techniques, especially in sloping areas (Anache et al. 2018; ÇEMGM 2017; Li and Fang 2016; Poesen 2018; Ziadat and Taimeh 2013). Effective measures for soil conservation can also use spatial patterns of plant cover to reduce sediment connectivity, and the relationships between hillslopes and sediment transfer in eroded channels (García-Ruiz et al. 2017). The following three examples present lessons learnt from the soil erosion problems and measures to address them in different settings of Chile, Turkey and the Central Asian countries.

Soil erosion by water is an important problem in Chile. National assessments conducted in 1979, which examined 46% of the continental surface of the country, concluded that very high levels of soil erosion affected 36% of the territory. The degree of soil erosion increases from south to north. The leading locations in Chile are the region of Coquimbo with 84% of eroded soils (Lat. 29°S, semi-arid climate), the region of Valparaíso with 57% of eroded soils (Lat. 33°S, Mediterranean climate) and the region of O’Higgins with 37% of eroded soils (Lat. 34°S, Mediterranean climate). The most important drivers of soil erosion are soil, slope, climate erosivity (i.e., precipitation, intensity, duration and frequency) due to a highly concentrated rainy season, and vegetation structure and cover. In the region of Coquimbo, goat and sheep overgrazing have aggravated the situation (CIREN 2010). Erosion rates reach up to 100 t ha–1 annually, having increased substantially over the last 50 years (Ellies 2000). About 10.4% of central Chile exhibits high erosion rates (greater than 1.1 t ha–1 annually) (Bonilla et al. 2010).

Over the last few decades there has been an increasing interest in the development of no-till (also called zero tillage) technologies to minimise soil disturbance, reduce the combustion of fossil fuels and increase soil organic matter. No-till, in conjunction with the adoption of strategic cover crops, has positively impacted soil biology with increases in soil organic matter. Early evaluations by Crovetto, (1998) showed that no-till application (after seven years) had doubled the biological activity indicators compared to traditional farming and even surpassed those found in pasture (grown for the previous 15 years). Besides erosion control, additional benefits are an increase of water-holding capacity and reduction in bulk density. Currently, the above no-till farm experiment has lasted for 40 years and continues to report benefits to soil health and sustainable production (Reicosky and Crovetto 2014). The influence of this iconic farm has resulted in the adoption of soil conservation practices – and especially no-till – in dryland areas of the Mediterranean climate region of central Chile (Martínez et al. 2011). Currently, it has been estimated that the area under no-till farming in Chile varies between 0.13 and 0.2 Mha (Acevedo and Silva 2003).

In Turkey, the amount of sediment recently released through erosion into seas was estimated to be 168 Mt yr-1 , which is considerably lower than the 500 Mt yr–1 that was estimated to be lost in the 1970s. The decrease in erosion rates is attributed to an increase in spatial extent of forests, rehabilitation of degraded forests, erosion control, prevention of overgrazing, and improvement in irrigation technologies. Soil conservation measures conducted in the Karapınar district, Turkey, exemplify these activities. The district is characterised by a semi-arid climate and annual average precipitation of 250–300 mm (Türkeş 2003; Türkeş and Tatlı 2011). In areas where vegetation was overgrazed or inappropriately tilled, the surface soil horizon was removed through erosion processes resulting in the creation of large drifting dunes that threatened settlements around Karapınar (Groneman 1968). Such dune movement had begun to affect the Karapınar settlement in 1956 (Kantarcı et al. 2011). Consequently, by the early 1960s, Karapınar town and nearby villages were confronted with the danger of abandonment due to out-migration in the early 1960s (Figure 3.11(1)). The reasons for increasing wind erosion in the Karapınar district can be summarised as follows: sandy material was mobilised following drying of the lake; hot and semi-arid climate conditions; overgrazing and use of pasture plants for fuel; excessive tillage; and strong prevailing winds.

Restoration and mitigation strategies were initiated in 1959, and today 4300 ha of land have been restored (Akay and Yildirim 2010) (Figure 3.11 (2)), using specific measures: (i) physical measures: construction of cane screens to decrease wind speed and prevent sand movement (Figure 3.11(3)); (ii) restoration of cover: increasing grass cover between screens using seeds collected from local pastures or the cultivation of rye (Secale sp.) and wheat grass (Agropyron elongatum) that are known to grow in arid and hot conditions; and (iii) afforestation: saplings obtained from nursery gardens were planted and grown between these screens. Main tree species selected were oleaster (Eleagnus sp.), acacia (Robinia pseudeaccacia), ash (Fraxinus sp.), elm (Ulmus sp.) and maple (Acer sp.) (Figure 3.11 (4)). Economic growth occurred after controlling erosion and new tree nurseries have been established with modern irrigation. Potential negative consequences through the excessive use of water can be mitigated through engagement with local stakeholders and transdisciplinary learning processes, as well as by restoring the traditional land uses in the semi-arid Konya closed basin (Akça et al. 2016).

Soil erosion is widely acknowledged to be a major form of degradation of Central Asian drylands, affecting a considerable share of croplands and rangelands. However, up-to-date information on the actual extent of eroded soils at the regional or country level is not available. The estimates compiled by Pender et al. (2009), based on the Central Asian Countries Initiative for Land Management (CACILM), indicate that about 0.8 Mha of the irrigated croplands were subject to high degree of soil erosion in Uzbekistan. In Turkmenistan, soil erosion was indicated to be occurring in about 0.7 Mha of irrigated land. In Kyrgyzstan, out of 1 Mha of irrigated land in the foothill zones, 0.76 Mha were subject to soil erosion by water, leading to losses in crop yields of 20–60% in these eroded soils. About 0.65 Mha of arable land were prone to soil erosion by wind (Mavlyanova et al. 2017). Soil erosion is widespread in rainfed and irrigated areas in Kazakhstan (Saparov 2014). About 5 Mha of rainfed croplands were subject to high levels of soil erosion (Pender et al. 2009). Soil erosion by water was indicated to be a major concern in sloping areas in Tajikistan (Pender et al. 2009).

The major causes of soil erosion in Central Asia are related to human factors, primarily excessive water use in irrigated areas (Gupta et al. 2009), deep ploughing and lack of maintenance of vegetative cover in rainfed areas (Suleimenov et al. 2014), and overgrazing in rangelands (Mirzabaev et al. 2016). Lack of good maintenance of watering infrastructure for migratory livestock grazing, and fragmentation of livestock herds led to overgrazing near villages, increasing the soil erosion by wind (Alimaev et al. 2008). Overgrazing in the rangeland areas of the region (e.g., particularly in Kyzylkum) contributes to dust storms, coming primarily from the Ustyurt Plateau, desertified areas of Amudarya and Syrdarya rivers’ deltas, the dried seabed of the Aral Sea (now called Aralkum), and the Caspian Sea (Issanova and Abuduwaili 2017; Xi and Sokolik 2015). Xi and Sokolik (2015) estimated that total dust emissions in Central Asia were 255.6 Mt in 2001, representing 10–17% of the global total.

Arid and semi-arid areas of China, including north-eastern, northern and north-western regions, cover an area of more than 509 Mha, with annual rainfall of below 450 mm. Over the past several centuries, more than 60% of the areas in arid and semi-arid regions were used as pastoral and agricultural lands. The coupled impacts of past climate change and human activity have caused desertification and dust storms to become a serious problem in the region (Xu et al. 2010). In 1958, the Chinese government recognised that desertification and dust storms jeopardised the livelihoods of nearly 200 million people, and afforestation programmes for combating desertification have been initiated since 1978. China is committed to go beyond the Land Degradation Neutrality objective, as indicated by the following programmes that have been implemented. The Chinese Government began the Three North’s Forest Shelterbelt programme in Northeast China, North China, and Northwest China, with the goal to combat desertification and to control dust storms by improving forest cover in arid and semi-arid regions. The project is implemented in three stages (1978–2000, 2001–2020 and 2021–2050). In addition, the Chinese government launched the Beijing and Tianjin Sandstorm Source Treatment Project (2001–2010), Returning Farmlands to Forest Project (2003–present), and the Returning Grazing Land to Grassland Project (2003–present) to combat desertification, and for adaptation and mitigation of climate change (State Forestry Administration of China 2015; Wang 2014; Wang et al. 2013).

The results of the fifth monitoring period (2010–2014) showed: (i) compared with 2009, the area of degraded land decreased by 12,120 km2 over a five-year period; (ii) in 2014, the average coverage of vegetation in the sand area was 18.33%, an increase of 0.7% compared with 17.63% in 2009, and the carbon sequestration increased by 8.5%; (iii) compared with 2009, the amount of wind erosion decreased by 33%, the average annual occurrence of sandstorms decreased by 20.3% in 2014; (iv) as of 2014, 203,700 km2 of degraded land were effectively managed, accounting for 38.4% of the 530,000 km2 of manageable desertified land; (v) the restoration of degraded land has created an annual output of 53.63 Mt of fresh and dried fruits, accounting for 33.9% of the total national annual output of fresh and dried fruits (State Forestry Administration of China 2015). This has become an important pillar for economic development and a high priority for peasants as a method to eradicate poverty (State Forestry Administration of China 2015).

After independence in 1962, the Algerian government initiated measures to replant forests destroyed by the war, and the steppes affected by desertification, among its top priorities (Belaaz 2003).

In 1972, the government invested in the Green Dam (Barrage vert) project. This was the first significant experiment to combat desertification, influence the local climate and decrease the aridity by restoring a barrier of trees. The Green Dam extends across arid and semi-arid zones between the isohyets 300 mm and 200 mm. It is a 3 Mha band of plantation running from east to west (Figure 3.12). It is over 1200 km long (from the Algerian–Moroccan border to the Algerian–Tunisian border) and has an average width of about 20 km. The soils in the area are shallow, low in organic matter and susceptible to erosion. The main objectives of the project were to conserve natural resources, improve the living conditions of local residents and avoid their exodus to urban areas. During the first four decades (1970–2000) the success rate was low (42%) due to lack of participation by the local population and the choice of species (Bensaid 1995).

The Green Dam did not have the desired effects. Despite tree-planting efforts, desertification intensified on the steppes, especially in south-western Algeria, due to the prolonged drought during the 1980s. Rainfall declined in the range from 18% to 27%, and the dry season has increased by two months in the last century (Belala et al. 2018). Livestock numbers in the Green Dam regions, mainly sheep, grew exponentially, leading to severe overgrazing, causing trampling and soil compaction, which greatly increased the risk of erosion. Wind erosion, very prevalent in the region, is due to climatic conditions and the strong anthropogenic action that reduced the vegetation cover. The action of the wind carries fine particles such as sands and clays and leaves on the soil surface a lag-gravel pavement, which is unproductive. Water erosion is largely due to torrential rains in the form of severe thunderstorms that disintegrate the bare soil surface from raindrop impact (Achite et al. 2016). The detached soil and nutrients are transported offsite via runoff, resulting in loss of fertility and water holding capacity. The risk of and severity of water erosion is a function of human land-use activities that increase soil loss through removal of vegetative cover. The National Soil Sensitivity to Erosion Map (Salamani et al. 2012) shows that more than 3 Mha of land in the steppe provinces are currently experiencing intense wind activity (Houyou et al. 2016) and that these areas are at particular risk of soil erosion. Mostephaoui et al. (2013), estimates that each year there is a loss of 7 t ha–1 of soils due to erosion. Nearly 0.6 Mha of land in the steppe zone are fully degraded without the possibility of biological recovery.

To combat the effects of erosion and desertification, the government has planned to relaunch the rehabilitation of the Green Dam by incorporating new concepts related to sustainable development, and adaptation to climate change. The experience of previous years has led to integrated rangeland management, improved tree and fodder shrub plantations and the development of water conservation techniques. Reforestation is carried out using several species, including fruit trees, to increase and diversify the sources of income for the population.

The evaluation of the Green Dam from 1972 to 2015 (Merdas et al. 2015) shows that 0.3 Mha of forest plantation have been planted, which represents 10% of the project area. Estimates of the success rate of reforestation vary considerably between 30% and 75%, depending on the region. Through demonstration, the Green Dam has inspired several African nations to work together to build a Great Green Wall to combat land degradation, mitigate climate change effects, loss of biodiversity and poverty in a region that stretches from Senegal to Djibouti (Sahara and Sahel Observatory (OSS) 2016) (Section 3.7.2.3).

The Great Green Wall is an initiative of the Heads of State and Government of the Sahelo-Saharan countries to mitigate and adapt to climate change, and to improve the food security of the Sahel and Saharan peoples (Sacande 2018; Mbow 2017). Launched in 2007, this regional project aims to restore Africa’s degraded arid landscapes, reduce the loss of biodiversity and support local communities to sustainable use of forests and rangelands. The Great Green Wall focuses on establishing plantations and neighbouring projects, covering a distance of 7775 km from Senegal on the Atlantic coast to Eritrea on the Red Sea coast, with a width of 15 km (Figure 3.13). The wall passes through Djibouti, Eritrea, Ethiopia, Sudan, Chad, Niger, Nigeria, Mali, Burkina Faso, Mauritania and Senegal.

The choice of woody and herbaceous species that will be used to restore degraded ecosystems is based on biophysical and socio-economic criteria, including socio-economic value (food, pastoral, commercial, energetic, medicinal, cultural); ecological importance (carbon sequestration, soil cover, water infiltration); and resilience to climate change and variability. The Pan-African Agency of the Great Green Wall (PAGGW) was created in 2010 under the auspices of the African Union and CEN-SAD to manage the project. The initiative is implemented at the level of each country by a national structure. A monitoring and evaluation system has been defined, allowing nations to measure outcomes and to propose the necessary adjustments.

In the past, reforestation programmes in the arid regions of the Sahel and North Africa that have been undertaken to stop desertification were poorly studied and cost a lot of money without significant success (Benjaminsen and Hiernaux 2019). Today, countries have changed their strategies and opted for rural development projects that can be more easily funded. Examples of scalable practices for land restoration include managing water bodies for livestock and crop production, and promoting fodder trees to reduce runoff (Mbow 2017).

The implementation of the initiative has already started in several countries. For example, the FAO’s Action Against Desertification project was restoring 18,000 hectares of land in 2018 through planting native tree species in Burkina Faso, Ethiopia, The Gambia, Niger, Nigeria and Senegal (Sacande 2018). Berrahmouni et al. (2016) estimated that 166 Mha can be restored in the Sahel, requiring the restoration of 10 Mha per year to achieve Land Degradation Neutrality targets by 2030. Despite these early implementation actions on the ground, the achievement of the planned targets is questionable, and will be challenging without significant additional funding.

The spread of invasive plants can be exacerbated by climate change (Bradley et al. 2010; Davis et al. 2000). In general, it is expected that the distribution of invasive plant species with high tolerance to drought or high temperatures may increase under most climate change scenarios (medium to high confidence) (Bradley et al. 2010; Settele et al. 2014; Scasta et al. 2015). Invasive plants are considered a major risk to native biodiversity and can disturb the nutrient dynamics and water balance in affected ecosystems (Ehrenfeld 2003). Compared to more humid regions, the number of species that succeed in invading dryland areas is low (Bradley et al. 2012), yet they have a considerable impact on biodiversity and ecosystem services (Le Maitre et al. 2015, 2011; Newton et al. 2011). Moreover, human activities in dryland areas are responsible for creating new invasion opportunities (Safriel et al. 2005).

Current drivers of species introductions include expanding global trade and travel, land degradation and changes in climate (Chytrý et al. 2012; Richardson et al. 2011; Seebens et al. 2018). For example, Davis et al. (2000) suggests that high rainfall variability promotes the success of alien plant species – as reported for semi-arid grasslands and Mediterranean-type ecosystems (Cassidy et al. 2004; Reynolds et al. 2004; Sala et al. 2006). Furthermore, Panda et al. (2018) demonstrated that many invasive species could withstand elevated temperature and moisture scarcity caused by climate change. Dukes et al. (2011) observed that the invasive plant yellow-star thistle (Centaurea solstitialis) grew six time larger under the elevated atmospheric CO2 expected in future climate change scenarios.

Climate change is likely to aggravate the problem as existing species continue to spread unabated and other species develop invasive characteristics (Hellmann et al. 2008). Although the effects of climate change on invasive species distributions have been relatively well explored, the greater impact on ecosystems is less well understood (Bradley et al. 2010; Eldridge et al. 2011).

Due to the time lag between the initial release of invasive species and their impact, the consequence of invasions is not immediately detected and may only be noticed centuries after introduction (Rouget et al. 2016). Climate change and invading species may act in concert (Bellard et al. 2013; Hellmann et al. 2008; Seebens et al. 2015). For example, invasion often changes the size and structure of fuel loads, which can lead to an increase in the frequency and intensity of fire (Evans et al. 2015). In areas where the climate is becoming warmer, an increase in the likelihood of suitable weather conditions for fire may promote invasive species, which in turn may lead to further desertification. Conversely, fire may promote plant invasions via several mechanisms (by reducing cover of competing vegetation, destroying native vegetation and clearing a path for invasive plants or creating favourable soil conditions) (Brooks et al. 2004; Grace et al. 2001; Keeley and Brennan 2012).

At a regional scale, Bellard et al. (2013) predicted increasing risk in Africa and Asia, with declining risk in Australia (Figure 3.14). This projection does not represent an exhaustive list of invasive alien species occurring in drylands.

A set of four case studies in Ethiopia, Mexico, the USA and Pakistan is presented below to describe the nuanced nature of invading plant species, their impact on drylands and their relationship with climate change.

The two invasive plants that inflict the heaviest damage to ecosystems, especially biodiversity, are the annual herbaceous weed, Parthenium hysterophorus (Asteraceae) also known as Congress weed; and the tree species, Prosopis juliflora (Fabaceae) also called Mesquite, both originating from the southwestern United States to Central/South America (Adkins and Shabbir 2014). Prosopis was introduced in the 1970s and has since spread rapidly. Prosopis, classified as the highest priority invader in Ethiopia, is threatening livestock production and challenging the sustainability of the pastoral systems. Parthenium is believed to have been introduced along with relief aid during the debilitating droughts of the early 1980s, and a recent study reported that it has spread into 32 out of 34 districts in Tigray, the northernmost region of Ethiopia (Teka 2016). A study by Etana et al. (2011) indicated that Parthenium caused a 69% decline in the density of herbaceous species in Awash National Park within a few years of introduction. In the presence of Parthenium, the growth and development of crops is suppressed due to its allelopathic properties. McConnachie et al. (2011) estimated a 28% crop loss across the country, including a 40–90% reduction in sorghum yield in eastern Ethiopia alone (Tamado et al. 2002). The weed is a substantial agricultural and natural resource problem and constitutes a significant health hazard (Fasil 2011). Parthenium causes acute allergic respiratory problems, skin dermatitis, and reportedly mutagenicity both in humans and livestock (Mekonnen 2017; Patel 2011). The eastern belt of Africa – including Ethiopia – presents a very suitable habitat, and the weed is expected to spread further in the region in the future (Mainali et al. 2015).

There is neither a comprehensive intervention plan nor a clear institutional mandate to deal with invasive weeds, however, there are fragmented efforts involving local communities even though they are clearly inadequate. The lessons learned, related to actions that have contributed to the current scenario, are several. First, lack of coordination and awareness – mesquite was introduced by development agencies as a drought-tolerant shade tree with little consideration of its invasive nature. If research and development institutions had been aware, a containment strategy could have been implemented early on. The second major lesson is the cost of inaction. When research and development organisations did sound the alarm, the warnings went largely unheeded, resulting in the spread and buildup of two of the worst invasive plant species in the world (Fasil 2011).

Buffelgrass (Cenchrus ciliaris L.), a native species from southern Asia and East Africa, was introduced into Texas and northern Mexico in the 1930s and 1940s, as it is highly productive in drought conditions (Cox et al. 1988; Rao et al. 1996). In the Sonoran desert of Mexico, the distribution of buffelgrass has increased exponentially, covering 1 Mha in Sonora State (Castellanos-Villegas et al. 2002). Furthermore, its potential distribution extended to 53% of Sonora State and 12% of semi-arid and arid ecosystems in Mexico (Arriaga et al. 2004). Buffelgrass has also been reported as an aggressive invader in Australia and the USA, resulting in altered fire cycles that enhance further spread of this plant and disrupt ecosystem processes (Marshall et al. 2012; Miller et al. 2010; Schlesinger et al. 2013).

Castellanos et al. (2016) reported that soil moisture was lower in the buffelgrass savannah cleared 35 years ago than in the native semi-arid shrubland, mainly during the summer. The ecohydrological changes induced by buffelgrass can therefore displace native plant species over the long term. Invasion by buffelgrass can also affect landscape productivity, as it is not as productive as native vegetation (Franklin and Molina-Freaner 2010). Incorporation of buffelgrass is considered a good management practice by producers and the government. For this reason, no remedial actions are undertaken.

Sagebrush ecosystems have declined from 25 Mha to 13 Mha since the late 1800s (Miller et al. 2011). A major cause is the introduction of non-native cheatgrass (Bromus tectorum), which is the most prolific invasive plant in the USA. Cheatgrass infests more than 10 Mha in the Great Basin and is expanding every year (Balch et al. 2013). It provides a fine-textured fuel that increases the intensity, frequency and spatial extent of fire (Balch et al. 2013). Historically, wildfire frequency was 60 to 110 years in Wyoming big sagebrush communities and has increased to five years following the introduction of cheatgrass (Balch et al. 2013; Pilliod et al. 2017).

The conversion of the sagebrush steppe biome to annual grassland with higher fire frequencies has severely impacted livestock producers, as grazing is not possible for a minimum of two years after fire. Furthermore, cheatgrass and wildfires reduce critical habitat for wildlife and negatively impact species richness and abundance – for example, the greater sage-grouse (Centocercus urophasianus) and pygmy rabbit (Brachylagus idahoensis) which are on the verge of being listed for federal protection (Crawford et al. 2004; Larrucea and Brussard 2008; Lockyer et al. 2015).

Attempts to reduce cheatgrass impacts through reseeding of both native and adapted introduced species have occurred for more than 60 years (Hull and Stewart 1949) with little success. Following fire, cheatgrass becomes dominant and recovery of native shrubs and grasses is improbable, particularly in relatively low-elevation sites with minimal annual precipitation (less than 200 mm yr–1) (Davies et al. 2012; Taylor et al. 2014). Current rehabilitation efforts emphasise the use of native and non-native perennial grasses, forbs and shrubs (Bureau of Land Management 2005). Recent literature suggests that these treatments are not consistently effective at displacing cheatgrass populations or re-establishing sage-grouse habitat, with success varying with elevation and precipitation (Arkle et al. 2014; Knutson et al. 2014). Proper post-fire grazing rest, season-of-use, stocking rates, and subsequent management are essential to restore resilient sagebrush ecosystems before they cross a threshold and become an annual grassland (Chambers et al. 2014; Miller et al. 2011; Pellant et al. 2004). Biological soil crust protection may be an effective measure to reduce cheatgrass germination, as biocrust disturbance has been shown to be a key factor promoting germination of non-native grasses (Hernandez and Sandquist 2011). Projections of increasing temperature (Abatzoglou and Kolden 2011), and observed reductions in and earlier melting of snowpack in the Great Basin region (Harpold and Brooks 2018; Mote et al. 2005) suggest that there is a need to understand current and past climatic variability as this will drive wildfire variability and invasions of annual grasses.

The alien plants invading local vegetation in Pakistan include Brossentia papyrifera (found in Islamabad Capital territory), Parthenium hysterophorus (found in Punjab and Khyber Pakhtunkhwa provinces), Prosopis juliflora (found all over Pakistan), Eucalyptus camaldulensis (found in Punjab and Sindh provinces), Salvinia (aquatic plant widely distributed in water bodies in Sindh), Cannabis sativa (found in Islamabad Capital Territory), Lantana camara and Xanthium strumarium (found in upper Punjab and Khyber Pakhtunkhwa provinces) (Khan et al. 2010; Qureshi et al. 2014). Most of these plants were introduced by the Forest Department decades ago for filling the gap between demand and supply of timber, fuelwood and fodder. These non-native plants have some uses but their disadvantages outweigh their benefits (Marwat et al. 2010; Rashid et al. 2014).

Besides being a source of biological pollution and a threat to biodiversity and habitat loss, the alien plants reduce the land value and cause huge losses to agricultural communities (Rashid et al. 2014). Brossentia papyrifera, commonly known as Paper Mulberry, is the root cause of inhalant pollen allergy for the residents of lush green Islamabad during spring. From February to April, the pollen allergy is at its peak, with symptoms of severe persistent coughing, difficulty in breathing, and wheezing. The pollen count, although variable at different times and days, can be as high as 55,000 m-3.

Early symptoms of the allergy include sneezing, itching in the eyes and skin, and blocked nose. With changing climate, the onset of disease is getting earlier, and pollen count is estimated to cross 55,000 m–3 (Rashid et al. 2014). About 45% of allergic patients in the twin cities of Islamabad and Rawalpindi showed positive sensitivity to the pollens (Marwat et al. 2010). Millions of rupees have been spent by the Capital Development Authority on pruning and cutting of Paper Mulberry trees but because of its regeneration capacity growth is regained rapidly (Rashid et al. 2014). Among other invading plants, Prosopis juliflora has allelopathic properties, and Eucalyptus is known to transpire huge amounts of water and deplete the soil of its nutrient elements (Qureshi et al. 2014).

The second great challenge after the Green Revolution in India was the low productivity in the rain-fed and semi-arid regions where land degredation and drought were serious concerns. In response to this challenge IWM projects were implemented over large areas in semi-arid biomes over the past few decades. IWM was meant to become a key factor in meeting a range of social development goals in many semi-arid rainfed agrarian landscapes in India (Bouma et al. 2007; Kerr et al. 2002). Over the years, watershed development has become the fulcrum of rural development, and has the potential to achieve the twin objectives of ecosystem restoration and livelihood assurance in the drylands of India (Joy et al. 2004).

Many reports indicate significant improvements in mitigation of drought impacts, raising crops and fodder, livestock productivity, expanding the availability of drinking water and increasing incomes as a result of IWM (Rao 2000), but in some cases overall the positive impact of the programme has been questioned and, except in a few cases, the performance has not lived up to expectations (Joy et al. 2004; JM Kerr et al. 2002). Comparisons of catchments with and without IWM projects using remotely sensed data have sometimes shown no significant enhancement of biomass, in part due to methodological challenges of space for time comparisons (Bhalla et al. 2013). The factors contributing to the successful cases were found to include effective participation of stakeholders in management (Rao 2000; Ratna Reddy et al. 2004).

Attribution of success in soil and water conservation measures was confounded by inadequate monitoring of rainfall variability and lack of catchment hydrologic indicators (Bhalla et al. 2013). Social and economic trade-offs included bias of benefits to downstream crop producers at the expense of pastoralists, women and upstream communities. This biased distribution of IWM benefits could potentially be addressed by compensation for environmental services between communities (Kerr et al. 2002). The successes in some areas also led to increased demand for water, especially groundwater, since there has been no corresponding social regulation of water use after improvement in water regime (Samuel et al. 2007). Policies and management did not ensure water allocation to sectors with the highest social and economic benefits (Batchelor et al. 2003). Limited field evidence of the positive impacts of rainwater harvesting at the local scale is available, but there are several potential negative impacts at the watershed scale (Glendenning et al. 2012). Furthermore, watershed projects are known to have led to more water scarcity, and higher expectations for irrigation water supply, further exacerbating water scarcity (Bharucha et al. 2014).

In summary, the mixed performance of IWM projects has been linked to several factors. These include: inequity in the distribution of benefits (Kerr et al. 2002); focus on institutional aspects rather than application of appropriate watershed techniques and functional aspects of watershed restoration (Joy et al. 2006; Vaidyanathan 2006); mismatch between scales of focus and those that are optimal for catchment processes (Kerr 2007); inconsistencies in criteria used to select watersheds for IWM projects (Bhalla et al. 2011); and in a few cases additional costs and inefficiencies of local non-governmental organisations (Chandrasekhar et al. 2006; Deshpande 2008). Enabling policy responses for improvement of IWM performance include: a greater emphasis on ecological restoration rather than civil engineering; sharper focus on sustainability of livelihoods than just conservation; adoption of ‘water justice’ as a normative goal and minimising externalities on non-stakeholder communities; rigorous independent biophysical monitoring, with feedback mechanisms and integration with larger schemes for food and ecological security, and maintenance of environmental flows for downstream areas (Bharucha et al. 2014; Calder et al. 2008; Joy et al. 2006). Successful adaptation of IWM to achieve land degradation neutrality would largely depend on how IWM creatively engages with dynamics of large-scale land use and hydrology under a changing climate, involvement of livelihoods and rural incomes in ecological restoration, regulation of groundwater use, and changing aspirations of rural population (robust evidence, high agreement) (O’Brien et al. 2004; Samuel et al. 2007; Samuel and Joy 2018).

Population growth, migration into Jordan and changes in climate have resulted in desertification of the Jordan Badia region. The Badia region covers more than 80% of the country’s area and receives less than 200 mm of rainfall per year, with some areas receiving less than 100 mm (Al-Tabini et al. 2012). Climate analysis has indicated a generally increasing dryness over the West Asia and Middle East region (AlSarmi and Washington 2011; Tanarhte et al. 2015), with reduction in average annual rainfall in Jordan’s Badia area (De Pauw et al. 2015). The incidence of extreme rainfall events has not declined over the region. Locally increased incidence of extreme events over the Mediterranean region has been proposed (Giannakopoulos et al. 2009).

The practice of intensive and localised livestock herding, in combination with deep ploughing and unproductive barley agriculture, are the main drivers of severe land degradation and depletion of the rangeland natural resources. This affected both the quantity and the diversity of vegetation as native plants with a high nutrition value were replaced with invasive species with low palatability and nutritional content (Abu-Zanat et al. 2004). The sparsely covered and crusted soils in Jordan’s Badia area have a low rainfall interception and infiltration rate, which leads to increased surface runoff and subsequent erosion and gullying, speeding up the drainage of rainwater from the watersheds, which can result in downstream flooding in Amman, Jordan (Oweis 2017).

In-depth understanding of the Badia’s rangeland status transition, coupled with sustainable rangeland management, are still subject to further investigation, development and adoption; a combination of all three is required to mitigate the ongoing degradation of the Middle Eastern rangeland ecosystems.

Oweis (2017) indicated that the cost of the fully automated Vallerani technique was approximately 32 USD ha-1. The total cost of the restoration package included the production, planting and maintenance of the shrub seedlings (11 USD ha–1). Tabieh et al. (2015) calculated a benefit-cost ratio (BCR) of above 1.5 for re-vegetation of degraded Badia areas through MIRWH and saltbush. However, costs vary based on the seedling’s costs and availability of trained labour.

Water harvesting is not a recent scientific advancement. Water harvesting is known to have been developed during the Bronze Age and was widely practiced in the Negev Desert during the Byzantine time period (1300–1600 years ago) (Fried et al. 2018; Stavi et al. 2017). Through construction of various structures made of packed clay and stone, water was either held on site in half-circular dam structures (hafir) that faced up-slope to capture runoff, or on terraces that slowed water allowing it to infiltrate and to be stored in the soil profile. Numerous other systems were designed to capture water in below-ground cisterns to be used later to provide water to livestock or for domestic use. Other water harvesting techniques divert runoff from hillslopes or wadis and spread the water in a systematic manner across playas and the toe-slope of a hillslope. These systems allow production of crops in areas with 100 mm of average annual precipitation by harvesting an additional 300+ mm of water (Beckers et al. 2013). Water harvesting is a proven technology to mitigate or adapt to climate change where precipitation may be reduced, and allow for small-scale crop and livestock production to continue supporting local needs.

Covering an area of 412,938 km2, the Limpopo River basin spans parts of Botswana, South Africa, Zimbabwe and Mozambique, eventually entering into the Mozambique Channel. It has been selected as a case study as it provides a clear illustration of the combined effect of desertification and climate change, and why IWM may be a crucial component of reducing exposure to climate change. It is predominantly a semi-arid area with an average annual rainfall of 400 mm (Mosase and Ahiablame 2018). Rainfall is both highly seasonal and variable, with the prominent impact of the El Niño/ La Niña phenomena and the Southern Oscillation leading to severe droughts (Jury 2016). It is also exposed to tropical cyclones that sweep in from the Mozambique Channel often leading to extensive casualties and the destruction of infrastructure (Christie and Hanlon 2001). Furthermore, there is good agreement across climate models that the region is going to become warmer and drier, with a change in the frequency of floods and droughts (Engelbrecht et al. 2011; Zhu and Ringler 2012). Seasonality is predicted to increase, which in turn may increase the frequency of flood events in an area that is already susceptible to flooding (Spaliviero et al. 2014).

A clear need exists to both address exposure to flood events as well as predicted decreases in water availability, which are already acute. Without the additional impact of climate change, the basin is rapidly reaching a point where all available water has been allocated to users (Kahinda et al. 2016; Zhu and Ringler 2012). The urgency of the situation was identified several decades ago (FAO 2004), with the countries of the basin recognising that responses are required at several levels, both in terms of system governance and the need to address land degradation.

Recent reviews of the governance and implementation of IWM within the basin recognise that an integrated approach is needed and that a robust institutional, legal, political, operational, technical and support environment is crucial (Alba et al. 2016; Gbetibouo et al. 2010; Machethe et al. 2004; Spaliviero et al. 2011; van der Zaag and Savenije 1999). Within the scope of emerging lessons, two principal ones emerge. The first is capacity and resource constraints at most levels. Limited capacity within Limpopo Watercourse Commission (LIMCOM) and national water management authorities constrains the implementation of IWM planning processes (Kahinda et al. 2016; Spaliviero et al. 2011). Whereas strategy development is often relatively well-funded and resourced through donor funding, long-term implementation is often limited due to competing priorities. The second is adequate representation of all parties in the process in order to address existing inequalities and ensure full integration of water management. For example, within Mozambique, significant strides have been made towards the decentralisation of river basin governance and IWM. Despite good progress, Alba et al. (2016) found that the newly implemented system may enforce existing inequalities as not all stakeholders, particularly smallholder farmers, are adequately represented in emerging water management structures and are often inhibited by financial and institutional constraints. Recognising economic and socio-political inequalities, and explicitly considering them to ensure the representation of all participants, can increase the chances of successful IWM implementation.