Numerous models have been developed across the last decade to assess biodiversity change (Melillo 1999, Alward and others 1999, Sala and others 2000, Clausenn and others 2003, Potting and Bakkes 2004). In 2004, several major global models merged in the creation of the GLOBIO 3.0 model, developed to meet the requirements of the CBD for a model suitable for simulating the rate and extent of biodiversity loss by estimating abundance of selected species, and to conduct projections of change in land use, infrastructure development, climate, and pollution. The individual components of the GLOBIO 3.0 model have been used to generate global scenarios (UNEP 2002b, CBD 2006), scenarios for the Arctic (Nellemann 2001, Ahlenius and others 2005), rainforest habitat for great apes in Asia and Africa (Nellemann and Newton 2003, Caldecott and Miles 2005, Butler 2003), and mountain regions (Blyth and others 2002, Nellemann 2005). The model results presented here use a definition of the desert biome described in Chapter 1, that is, an area of particular aridity, eco-regional and land-cover attributes.

By using the SRES scenarios (see Box 6.1 for details), which have previously been used to describe the implications of different socioeconomic developments for climate change, four different biodiversity scenarios have been developed for the desert biome. In this report, we present the results for biodiversity of the A2 scenario, or "regionally-defined market force" scenario. This scenario assumes that market forces will continue to be the main drivers of natural resource use, but that globalization is likely to reach a limit, giving way to renewed emphasis on local economies. We run the simulations for this scenario on the desert biome. The results of the other alternative scenarios are presented in parentheses as a range to provide an indication of the level of uncertainty. However, the A2 scenarios must not be considered a worst-case situation, but as the scenario closest to the trends of development as we have known them in the previous decades with regard to land degradation and subsequent impact on biodiversity and human livelihoods. It is also important to emphasize that because of the time lag for measures to control environmental degradation to take effect, all scenarios, even those with outstanding new efforts, will slow but not stop the rate of biodiversity loss.

Precipitation and temperature patterns likely to change

According to the SRES A2 scenario, great changes in both rainfall and temperature (and subsequent growth conditions) may take place across the world's deserts (Figure 6.3). The impacts will be highly variable from one region to the next, but they are likely to be felt the hardest in desert margins and in montane areas, as these are where the primary arid rangelands are located. Because deserts are driven by climatic pulses more than by average conditions, even moderate changes in precipitation and temperature may create severe impacts by shifting the intensity and frequency of extreme periods, and with perhaps catastrophic effects on the viability of plants, animals, and human livelihoods.

Land use intensifies in desert margins

Agricultural development, including irrigation, croplands, and grazing, is generally concentrated in oases, rangelands at desert margins, or in the lower slopes of montane desert areas. Great changes have taken place on the margins of virtually all desert areas in the world, particularly by grazing, over the last 150 years (Goldewijk 2001, Loreau and others 2001, Tilman and others 2001). According to the scenario, while expansion of croplands into deserts will be limited - except where fueled by irrigation - grazing by livestock and cutting of firewood will continue to increase inside deserts in montane areas, as well as on the desert margins (Figure 6.6).

Piecemeal development of sky-islands and desert margins

The changes and effects of land use both on the desert margins and inside the desert regions are deeply influenced by, and reflected in, piecemeal development of transportation networks, which are necessary for accessing, developing, and transporting people, goods and services, and for agricultural and livestock expansion (Leinbach 1995). Infrastructure development has been shown to disrupt the physical environment, alter the chemical environment, impact species relationships, accelerate introduction of invasive species, modify animal behavior and change land use near developed roads (Andrews 1990, Forman and Alexander 1998, Trombulak and Frissell 2000, Nellemann 2001). Desert wilderness areas (any area located more than 5 km from any infrastructure), are projected to decline from 59 per cent of the total desert area in 2005 to a low 31 per cent by 2050 (range 31-44 %), suggesting a relative loss of nearly half of the remaining intact wilderness within a few decades. This decline will primarily affect the more productive areas in desert margins and in montane areas, while the wilderness areas that remain will be primarily confined to barren areas with very low biodiversity, and where human settlements or development are not possible (SRES A2; Figure 6.7). Populations of desert bighorn sheep (Ovis canadensis), desert tortoises (Gopherus agassizii), and many species of birds have been shown to be very sensitive to fragmentation of habitat by roads (Bleich and others 1990, Edwards and others 2004, Epps and others 2005, Gutzwiller and Barrow 2003). The same applies to many species of antelope, which are vulnerable to poaching concentrated along road corridors (Nellemann 2005), or Asian houbara bustards (Chlamydotis macqueenii; Bekenov and others 1998, Spalton and others 1999, Combreau and others 2001, Mesochina and others 2003).

Desert sky-islands and wetlands in alarming decline

While the wilderness areas in hot deserts decline in the model by up to about 0.8 per cent every year as a result of human development and disturbance, the change in the desert margins is much greater (Figure 6.8). Here, relatively pristine natural rangelands may decline by 1.9 per cent annually. Wetlands are at even greater risk, as they are being drained for irrigation and agricultural expansion (see Chapter 5). Of greatest risk are the few patches of forest and woodlands associated with desert montane areas and the relatively moister desert margins, or riparian habitats next to settlements (Bleich and others 1990, Bekenov and others 1998, Spalton and others 1999, Combreau and others 2001, Mesochina and others 2003, Gutzwiller and Barrow 2003, Zhao and others 2004, Epps and others 2005, Nellemann 2005).

These areas are important not only for biodiversity, but are targeted because of water resources, potential pastures for livestock, and are also subject to cutting for firewood - a scarce resource in drylands. Pristine woodlands in deserts, such as montane habitats, may decline by up to 3.5 per cent annually, especially at lower elevations. This is particularly alarming, as the vegetation may be essential for reducing erosion, and logging may increase sediment loads in rivers, reduce water quality, and increase the risk and severity of flash floods (Nellemann 2005). Currently, deciduous forests and needle-leaf forests cover only 0.13 and 0.68 per cent of the desert biome in isolated sky-island patches respectively, but represent major biodiversity hotspots at risk, as they are frequently targeted for development. Historically, montane areas have been important for cross-desert transport and settlements because they constitute important water sources (see chapter 4). Wetlands in deserts are even rarer, occupying less than 0.01 per cent of the biome.

They are also projected to decline at fast rates, mainly due to drainage for irrigation and cropland development. All these habitats form hotspots of biodiversity and exhibit a very patchy distribution (Hernández and Bárcenas 1996, Riemann and Ezcurra 2005), restricted to mountainous regions (Hernández and others 2001), or to riparian zones, wadis, and in oases (Zhao and others 2004). The vulnerability of these habitats is mostly due to their isolation and fragmentation, lack of opportunity for migration of the biota when conditions change, limited extent and high endemism, and locallyrestricted species that are particularly vulnerable to change (Ezcurra and others 2001).

Apart from providing important resources for livestock grazing, montane habitats are also crucial for water supply to the surrounding deserts (Table 6.1); cutting of their forests greatly diminishes the ability of the mountains to regulate water flow. Hence, casualties are not uncommon downstream from flash floods exacerbated by unsustainable land practices (Nellemann 2005). Unfortunately, only a fraction of these montane habitats is protected.

Assessing biodiversity loss in deserts

During the 6th meeting of the Conference of Parties of the CBD, the parties committed themselves to "achieve by 2010 a significant reduction of the current rate of biodiversity loss at the global, regional and national levels" (CBD 2002). Until recently, there was little quantitative data available on recent changes in species abundance, and most studies relied extensively on expert or qualitative judgments (Leemans 2000, Sala and others 2000). Species richness was found to be an insufficient indicator. On the one hand, it is hard to monitor the number of species in an area, but, more importantly, it may sometimes increase as original species are gradually replaced by new human-introduced invasives. Consequently the CBD has chosen a limited set of indicators to track this degradation process, selecting, among others, the "change in abundance of selected species" (CBD 2004). The GLOBIO 3.0 model was developed specifically to estimate this indicator. Biodiversity loss is here expressed as the percent age of original species abundance as found in undisturbed controls or in information about the diversity in the original state of the land use category in question (Nellemann 2005, Scholes and Biggs 2005).

Change in local biodiversity in the world's deserts, 1700-2050 In desert regions, examples of human impacts include fragmentation of wildlife habitats by roads and dams as illustrated above (Epps and others 2005), illegal exploitation of cactoids and reptiles (Goode and others 2005), poaching of wildlife (Spalton and others 1999, Combreau and others 2001, Mesochina and others 2003), as well as land degradation associated with human activity in oases, wadis, and sky islands (Zhao and others 2004). Introduction of new invasive species, like the African buffel grass (Pennisetum ciliare) introduced in Sonora to improve rangelands for cattle production, also includes major new threats to desert biodiversity (Franklin and others 2006). Currently, the desert biome holds an average abundance of original species of 68 per cent. Most of it is concentrated in hotspots or in transition zones between arid rangelands and true deserts. In 1700, mean abundance of original species was approximately 93 per cent (range 89-96%), dropping to 87 per cent in 1900 (range 83-91%). Given a proportional decline in abundance with either (a) population growth or (b) change in cropland, the rate of loss in original species abundance has been about 0.17 per cent per year (range 0.13-0.21 %) in the last century in deserts. This compares to the decline of 0.8-2.4 per cent of intact wilderness ecosystems per year. Declines are greatest in desert margins and mountainous areas within deserts. Losses are also pronounced in coastal areas with high population density. Future losses of biodiversity in deserts Scenarios of change show that the rate of biodiversity loss in deserts may as much as double in the coming decades. These results are fairly similar compared to the global regime (CBD 2006). All four scenarios project a further decline in mean original species abundance from about 65 per cent in deserts in 2000 to a mean of 62.8 per cent by 2030 (range 60-65 %) and 58.3 per cent by 2050 (range 53-62 %; Figure 6.9).

Over a period of 50 years, the current global desert species abundance may thus drop by as much as 15 per cent - a dramatic decline given the relatively short timespan. The projected decline in biodiversity varies greatly among the scenarios. Remarkably, even a slowing of the rate of biodiversity loss to that of the mid 20th century would still mean a continued decline in the abundance of wildlife in desert regions. As for the scenarios in which the effect of land use and infrastructure development is modelled, the degree of decline varies greatly among and within the regions. Areas at desert biome boundaries or at higher elevations, such as in sky islands, are particularly prone and sensitive to change.

Ranking of pressures to biodiversity loss in deserts

Agriculture and human land use accounted for 41 per cent of the biodiversity loss by the year 2000. Fragmentation associated with infrastructure comes in at a close second (40 per cent). The relative share of the different factors varies among the scenarios, with climate change being the only one increasing in share for all four scenarios, from 6 per cent in 2000 to up to 14 per cent by 2050 (Thomas and others 2004), compared to a range of shares of 37-44 per cent for agriculture and 33-45 per cent for infrastructure. In deserts, infrastructure appears to play a major role in biodiversity losses, simply because it accelerates and facilitates human access to scattered and patchy hotspots of biodiversity where water is available, and because it increases fragmentation, which has been shown to have cascading adverse effects on ecosystems.