IPCC Fourth Assessment Report: Climate Change 2007
Climate Change 2007: Working Group II: Impacts, Adaptation and Vulnerability

4.4.2 Deserts

Properties, goods and services

One of the largest terrestrial biomes, deserts cover 27.7 Mkm2, comprising extra-polar regions with mean annual precipitation <250 mm and an unfavourable precipitation to potential evaporation ratio (Nicholson, 2002; Warner, 2004; Reid et al., 2005). Deserts support on the order of 10 people per km2, in sparse populations with among the lowest gross domestic product (GDP) of all ecoregions (Reid et al., 2005). Recent estimates suggest that between 10 and 20% of deserts and drylands are degraded due to an imbalance between demand for and supply of ecosystem services (Adeel et al., 2005). Critical provisioning goods and services include wild food sources, forage and rangeland grazing, fuel, building materials, and water for humans and livestock, for irrigation and for sanitation, and genetic resources, especially of arid-adapted species (Adeel et al., 2005; Hassan et al., 2005). Regulating services include air quality, atmosphere composition and climate regulation (Hassan et al., 2005), especially through wind-blown dust and desert albedo influences on regional rainfall, and biogeochemistry of remote terrestrial and marine ecosystems (Warner, 2004).

Key vulnerabilities

The TAR noted several vulnerabilities in drylands (Gitay et al., 2001, p. 239) but chiefly that human overuse and land degradation, exacerbated by an overall lack of infrastructure and investment in resource management, would be very likely to overwhelm climate change impacts, with the exception of impacts of increased dry and wet extremes due to ENSO frequency increase, and negative impacts of projected warming and drying in high biodiversity regions. On the other hand, evidence for region-specific increases in productivity and even community compositional change due to rising atmospheric CO2 was reported, with associated increased biomass and soil organic matter. Overall impacts of elevated CO2 were reported as comparable, though usually opposite in sign, to climate change projections. Since the TAR, further work shows that desert biodiversity is likely to be vulnerable to climate change (Reid et al., 2005), with winter-rainfall desert vegetation and plant and animal species especially vulnerable to drier and warmer conditions (Lenihan et al., 2003; Simmons et al., 2004; Musil et al., 2005; Malcolm et al., 2006), and continental deserts vulnerable to desiccation and even soil mobilisation, especially with human land-use pressures (Thomas and Leason, 2005). However, the potentially positive impact of rising atmospheric CO2 remains a significant uncertainty, especially because it is likely to increase plant productivity, particularly of C3 plants (Thuiller et al., 2006b) and, together with rainfall change, could even induce wildfires (Bachelet et al., 2001; Hardy, 2003; Duraiappah et al., 2005). The uncertain impact of elevated CO2 on vegetation productivity and biogeochemical cycling in deserts is an important source of contrasting projections of impacts and vulnerability for different desert regions and vegetation types. Climate change and direct human land-use pressure are likely to have synergistic impacts on desert ecosystems and species that may be offset, at least partly, by vegetation productivity and carbon sequestration gains due to rising atmospheric CO2. The net effect of these trends is very likely to be region-specific.


Deserts are likely to experience more episodic climate events, and interannual variability may increase in future, though there is substantial disagreement between GCM projections and across different regions (Smith et al., 2000; Duraiappah et al., 2005). Continental deserts could experience more severe, persistent droughts (Lioubimtseva and Adams, 2004; Schwinning and Sala, 2004). Vulnerability to desertification will be enhanced due to the indicated increase in the incidence of severe drought globally (Burke et al., 2006). In the Americas, temperate deserts are projected to expand substantially under doubled CO2 climate scenarios (Lauenroth et al., 2004). However, dry-spell duration and warming trend effects on vegetation productivity may be at least partly offset by rising atmospheric CO2 effects on plants (Bachelet et al., 2001; Thuiller et al., 2006b), leading to sometimes contrasting projections for deserts that are based on different modelling techniques that either incorporate or ignore CO2-fertilisation effects.

Elevated CO2 has been projected to have significant potential impacts on plant growth and productivity in drylands (Lioubimtseva and Adams, 2004). This projection has been confirmed for cool desert shrub species (Hamerlynck et al., 2002), and both desert shrubs and invasive (but not indigenous) grasses in wet years only (Smith et al., 2000). On the whole, evidence for CO2-fertilisation effects in deserts is conflicting, and species-specific (Lioubimtseva and Adams, 2004; Morgan et al., 2004). In the south-western USA the total area covered by deserts may decline by up to 60% if CO2-fertilisation effects are realised (Bachelet et al., 2001). Limited direct impacts of atmospheric CO2 on nitrogen-fixation have been found in soil biological crusts (Billings et al., 2003), but soil microbial activity beneath shrubs has been observed to increase, thus reducing plant-available nitrogen (Billings et al., 2002).

Soil vulnerability to climate change is indicated by shallow desert substrates with high soluble salts and the slow recolonisation of disturbed soil surfaces by different algae components (Evans and Belnap, 1999; Johansen, 2001; Duraiappah et al., 2005). Very low biomass (a drop below a 14% cover threshold) is very likely to make the Kalahari desert dune system in southern Africa susceptible to aeolian erosion (Thomas and Leason, 2005) and, with regional warming of between 2.5 and 3.5°C, most dune fields could be reactivated by 2100 (Thomas and Leason, 2005). Increased dust flux may increase aridity and suppress rainfall outside deserts, with opposite effects under wetting scenarios (Bachelet et al., 2001; Hardy, 2003; Prospero and Lamb, 2003; Lioubimtseva and Adams, 2004), leading to indirect effects on the vulnerability of remote regions to climate change. About one-third of the Sahel was projected to aridify with warming of 1.5 to 2°C by about 2050, with a general equatorward shift of vegetation zones (van den Born et al., 2004; Box 4.2). Alternative climate scenarios show less pronounced changes (van den Born et al., 2004).

Box 4.2. Vegetation response to rainfall variability in the Sahel

The Sahel falls roughly between the 100-200 mm/year (northern boundary) and 400-600 mm/year rainfall isohyets (southern boundary), and supports dry savanna vegetation forming transition zones with the Sahara and humid tropical savanna (Nicholson, 2000; Hiernaux and Turner, 2002; Anyamba and Tucker, 2005). These transition zones have historically fluctuated in response to rainfall changes (Hiernaux and Turner, 2002), in the clearest example of multi-decadal variability measured during the past century (Hulme, 2001). Ecosystem responses to past rainfall variability in the Sahel are potentially useful as an analogue of future climate change impacts, in the light of projections that extreme drought-affected terrestrial areas will increase from 1% to about 30% globally by the 2090s (Burke et al., 2006).

During the mid-Holocene, conditions supporting mesic vegetation and abundant wildlife deteriorated rapidly (ECF, 2004; Foley et al., 2003), highlighting the Sahel’s sensitivity to forcing effects. The Sahel has shown the largest negative trends in annual rainfall observed globally in the past century, though these reversed somewhat after the late 1970s (Trenberth et al., 2007). Since about 1900, multi-decadal-scale rainfall variability persisted, with drying trends between around 1930-1950 and 1960-1985 (Hulme, 2001; Nicholson, 2001). Conditions apparently improved between 1950 and 1960, with limited evidence suggesting increased human and livestock numbers (Reij et al., 2005). Severe drought prevailed in the early 1980s (Hulme, 2001; Trenberth et al., 2007), and groundwater levels declined, species-specific woody plant mortality increased (mainly of smaller plants), and even dominant perennial C4 grasses with high water-use efficiency declined. Exposed soil caused increased atmospheric dust loads (Nicholson, 2000, 2001). These events stimulated the concept of desertification and subsequent debates on its causes (Herrmann and Hutchinson, 2005).

The persistence of drought during the latter part of the 20th century prompted suggestions that land-cover change had exerted a positive feedback to reinforce drought conditions, but the modelled vegetation change necessary to induce this effect does not reflect reality (Hulme, 2001). During relatively wet periods (Nicholson et al., 2000; Anyamba and Tucker, 2005; Trenberth et al., 2007) spatially variable regeneration in both the herbaceous and the woody layer have been observed (Gonzalez, 2001; Rasmussen et al., 2001; Hiernaux and Turner, 2002). Remote sensing shows the resilience of Sahelian vegetation to drought, with no directional change in either desert transition zone position or vegetation cover (Nicholson et al., 1998). Sahel green-up between the years 1982 and 1998 (Prince et al., 1998; Hickler et al., 2005) and between 1994 and 2003 (Anyamba and Tucker, 2005) has been noted, but this interpretation has recently been challenged (Hein and Ridder, 2006).

Drivers of Sahel vegetation change remain uncertain (Hutchinson et al., 2005), especially because the correlation between rainfall and Normalised Difference Vegetation Index (NDVI) appear weak, signalling that greening cannot be fully explained by increasing rainfall (Olsson et al., 2005), and greening may not comprise a return to the initial species composition, cover and surface soil conditions (Warren, 2005). Inconclusive interpretations of vegetation dynamics in the Sahel may reflect complex combined effects of human land use and climate variability on arid environments (Rasmussen et al., 2001). It is far from clear how the interactive effect of climate change, land-use activities and rising CO2 will influence the Sahel in future. Green-up, or desert amelioration (Figure 4.3, vegetation class 4) due to rising CO2 and enhanced water-use efficiency (as observed by Herrmann et al., 2005) may accrue only in wet years (Morgan et al., 2004).

Episodic wet periods may increase vulnerability to invasive alien species and subsequent fire outbreaks and this, combined with land overuse, will increase vulnerability to degradation and desertification (Dukes and Mooney, 1999; Dube and Pickup, 2001; Holmgren and Scheffer, 2001; Brooks et al., 2004; Geist and Lambin, 2004; Lioubimtseva and Adams, 2004). Wet spells with elevated humidity and warmer temperatures will increase the prevalence of plant diseases (Harvell et al., 2002).

Desert biodiversity is likely to be vulnerable to climate change (Reid et al., 2005), especially in so-called ‘biodiversity hotspots’ (Myers et al., 2000). In the Succulent Karoo biome of South Africa, 2,800 plant species face potential extinction as bioclimatically suitable habitat is reduced by 80% with a global warming of 1.5-2.7°C above pre-industrial levels (see Table 4.1). Daytime in situ warming experiments suggest high vulnerability of endemic succulent (see Glossary) growth forms of the Succulent Karoo to high-end warming scenarios for 2100 (mean 5.5°C above current ambient temperatures), inducing appreciable mortality in some (but not all) succulent species tested within only a few months (Musil et al., 2005). Desert species that depend on rainfall events to initiate breeding, such as resident birds, and migratory birds whose routes cross deserts, will be severely affected (Dukes and Mooney, 1999; Hardy, 2003; Box 4.5). The Mountain Wheatear in South Africa was projected to lose 51% of its bioclimatic range by 2050 under an SRES A2 emissions scenario (Simmons et al., 2004). In contrast, desert reptile species could be favoured by warming, depending on rainfall scenario (Currie, 2001).