Global conservation efforts have made substantial gains in the past 20 years, aided by numerous large-scale scientific examinations and recommendations1,2,3,4. Much scientific and conservation efforts have focused on the megadiverse tropics5,6 while neglecting other important biomes7,8. Drylands encompass nearly half of the earth’s terrestrial surface, support vital ecosystems and endemic species, and provide key ecosystem services to a third of the global human population9,10. However, drylands are mostly absent from many dedicated global conservation initiatives11,12.

Drylands are especially sensitive to increasing human land-use pressures through land conversion due to extensive agriculture and alternative energy sources, overgrazing, invasive species and climate change—pressures that are expected to intensify in the coming decades13,14,15,16,17. Human pressures in drylands may result in severe land degradation and diminished overall productivity, threatening dryland ecosystems and biodiversity18. It is estimated that up to 20% of drylands are already degraded or at risk9,18, including large increases in bare ground and lost vegetation cover15 and considerable water declines in natural lakes, reservoirs19 and groundwater20. Drylands are also sensitive to the introduction and spread of non-native species supported by exogeneous resources due to agricultural and other human land uses in otherwise low-productivity desert ecosystems, with cascading ecological effects21,22,23. Greening initiatives and afforestation (for example, the Great Green Wall Initiative) have the potential to exacerbate these drivers through habitat and resource changes24,25,26. Dryland species are also among the most vulnerable to projected global climate changes resulting in extreme heat events and increased aridity that can overwhelm species already at their physiological tolerance limits to heat and water stress27,28,29. Climate change impacts may further be exacerbated in drylands by reduced opportunities for species to adapt and survive due to accelerated habitat loss and fragmentation13.

Consequently, there is an urgent need for a systematic approach to evaluate current and future threats to drylands and their vertebrate biodiversity. Such an approach will have important implications for guiding land management and conservation strategies in drylands by identifying broad-scale conservation priorities30 and increasing conservation targets across all ecosystems. Here we assess the degree of protected area coverage in drylands and dryland subtypes compared to non-drylands, the current status of anthropogenic threats to dryland vertebrate biodiversity—highlighting important protection gaps—and the impact that projected future human land-use pressures will have on drylands under different climate and socioeconomic pathways.

Global drylands

We evaluated drylands as classified by UNEP-WCMC (United Nations Environment World Conservation Monitoring Centre)31. Drylands are categorized using an aridity index (the ratio of annual precipitation to potential evapotranspiration), with values below 0.65. Drylands are further divided into hyperarid, arid, semiarid and dry subhumid subtypes (Fig. 1). This drylands designation dataset is commonly referred to by the Convention of Biological Diversity ( and the United Nations Convention to Combat Desertification ( for policy and management goals and often cited in the scientific literature (for example, ref. 10). Drylands cover about 42% of earth’s terrestrial surface and encompass diverse and unique regions globally (Fig. 1). Most dryland regions are found in Africa and Asia, which combined encompass 64% of global drylands and also comprise the largest areas of hyperarid and arid subtypes (Table 1). Drylands are mostly inclusive of the deserts and xeric shrublands biome1,2 and also comprise large proportions of other biomes and diverse habitats9 across continents. These range from subhumid zones (for example, tropical and subtropical grasslands, savannas and shrublands) especially in regions of Africa and Australia (Fig. 1 and Extended Data Fig. 1) to Mediterranean regions of Europe, temperate zones (for example, temperate grasslands, savannas and shrublands) especially in Asia and North America and small portions of sub-polar regions (for example, boreal forests/taiga and temperate conifer forests).

Fig. 1: Global drylands and subtypes (UNEP-WCMC)31.
figure 1

Circles show the relative terrestrial proportion of dryland subtypes globally and by continent.

Source data

Table 1 Geographic extent of drylands by subtype and continent

Protected areas in drylands

Drylands are considerably less covered by the global network of protected areas compared to non-drylands (Fig. 2a). Using the World Database on Protected Areas (WDPA)32 and OpenStreetMap33, we show that only ~12% of total drylands are covered by protected areas (considering all International Union for Conservation of Nature (IUCN) categories and uncategorized protected areas)—well below global conservation targets to be achieved by 2030 (that is, 30% coverage of ecologically representative areas). By contrast, non-dryland regions are better protected with ~21% coverage (Fig. 2a). Some of the largest protected areas in the world are in hyperarid deserts (for example, Uruq Bani Ma’arid in Saudi Arabia), presumably due to low human populations and few competing land uses34,35. However, overall the most arid dryland regions, which are home to unique and diverse plant and animal species9,13, are the least protected (Fig. 2a and Extended Data Fig. 2). For example, dry subhumid regions are 14.4% protected compared to hyperarid regions with only 6.7% protection. Furthermore, different biomes within drylands receive different levels of protection (Extended Data Fig. 3). For example, grasslands and savannahs are among the most biodiverse dryland ecosystems, vulnerable to over-grazing36 and other impacts, and least protected30. Considering protected areas managed mainly for science, wilderness protection and habitat and species conservation (that is, IUCN categories I–IV)37—a considerably smaller proportion of drylands are protected. Less than 5% of all dryland subtypes are found in protected areas designated as categories I–IV (Fig. 2a). These patterns remain consistent across continents (Fig. 2b). Such gaps in protected area coverage of drylands have important implications in Africa and Asia, which contain most global drylands (Tables 1 and 2). In Africa and Asia, protected areas including community and indigenous conserved areas without an IUCN category may be under-reported38, which might actively support effective biodiversity conservation39. Conversely, IUCN-designated protected areas may remain unprotected due to poor enforcement1. Nevertheless, all continents have a lower coverage of protected areas in categories I–IV (Fig. 2b). For example, in Australian drylands protection decreases from 20% to 6% coverage under IUCN categories I–IV and to less than 7% protection of North American drylands in IUCN categories I–IV. These findings are concerning, as protected areas permitting multiple land uses and resource extraction in drylands (that is, IUCN categories V–VI and uncategorized) might be particularly ineffective in biodiversity conservation due to the sensitivity of fragile dryland ecosystems and species to anthropogenic activities and threats6,40,41 (see ‘Threats to dryland biodiversity and conservation gaps’ and ‘Current and future land-use change scenarios’). Altogether, in drylands we find a disproportionate degree of low protection coverage overall, under-representation in the coverage of important dryland subtypes and biomes, and a low proportion of protected areas designated for biodiversity conservation. These factors are likely to have important implications for the conservation of dryland species and habitats.

Fig. 2: Protected areas in drylands.
figure 2

a, Proportion of protected areas by dryland subtype compared to total non-drylands. b, Proportion of protected areas in drylands by continent compared to non-drylands. Coloured bars, IUCN categories I–IV; white bars, IUCN categories V and VI and uncategorized (that is, ‘not applicable’, ‘not assigned’ and ‘not reported’).

Source data

Table 2 Protected areas in drylands by subtype and continent

Threats to dryland biodiversity and conservation gaps

Drylands contain a rich diversity of unique and endemic plant and animal species as a result of wide-ranging adaptations to extreme conditions over heterogeneous habitats9,13,42. We evaluated the distribution and threat status of terrestrial vertebrate species—amphibian, bird, mammal and reptile species with 50% or more of their global ranges in drylands. We find that the vast majority of dryland species have less than 10% of their ranges protected (considering IUCN protected area categories I–IV)37 in drylands (Fig. 3 and Supplementary Data 1), which is likely insufficient habitat for maintaining viable populations14,43. Worse still, 30% of amphibians, 7% of birds, 16% of mammals and 27% of reptiles have no overlap with protected areas (that is, ‘gap species’; Fig. 3, inner circles). Of these species with zero protection in drylands, 224 amphibian, 73 bird, 144 mammal and 720 reptile species are endemic to drylands (that is, species with more than 99% of their global ranges in drylands; Supplementary Data 1). Reptiles are the most diverse of the vertebrate groups found in drylands globally (total n = 3,589)—80% have less than 10% of their dryland distributions protected (Fig. 3d, inner circles), 53% are endemic to drylands and 13% are listed as threatened by the IUCN Red List (categories Critically Endangered (CR), Endangered (EN) and Vulnerable (VU); Fig. 3d, outer circles). Australia is a hotspot of reptile richness in drylands, with 726 species, 58% of which are endemic to drylands; 75% of all species have less than 10% of their distributions protected, and 5% are threatened (Extended Data Fig. 4). Other regions also emerge as important for dryland biodiversity due to their species richness. For example, South America is a hotspot of dryland biodiversity across all vertebrate groups (Extended Data Fig. 4) especially considering its relatively small area (9% of global drylands; Table 1). African and Asian drylands combined, matching their vast scale, host the highest richness of vertebrate species—45% (n = 3,627) of global dryland vertebrate species (Supplementary Data 1).

Fig. 3: Range size protection and threat status of terrestrial vertebrate species in drylands.
figure 3

ad, Amphibian (a), bird (b), mammal (c) and reptile (d) species in drylands (≥50% global range in drylands). Inner circles show proportion of species with range size protected (IUCN categories I–IV): orange, 0% range protected; blue, 0–10% range protected; green, 10–100% range protected. Outer circles show proportion of IUCN threatened species: dark orange, threatened (IUCN categories CR, EN and VU); dark green, non-threatened (IUCN categories Near Threatened (NT) and Least Concern (LC)); grey, data deficient (IUCN category Data Deficient (DD)); n, number of species; parentheses, number of endemic species (>99% range in drylands).

Source data

In addition, we evaluated the protection of dryland species by setting more demanding representation targets of protected area coverage for species with restricted ranges44. We set a 100% protection target for species with ranges smaller than 1,000 km2 in drylands and a 10–100% protection target for species with intermediate and widespread ranges above 1,000 km2. We find that most narrow-ranging species in drylands are inadequately protected (Fig. 4 and Supplementary Data 2). For example, 33% of amphibian species in drylands have very narrow ranges (n = 331); of these only 24 species have adequate protection, while 193 species have zero protection. Similarly, 18% of reptiles have very narrow ranges (n = 629); of these only 36 species have adequate protection, whereas 431 species have zero protection. Narrow-ranging birds (n = 65) and mammals (n = 96) are less common; however, of these only 2 birds and 3 mammals are adequately protected. Such protection gaps are concerning, as narrow-ranging species are in greatest need of protection due to various threats and are consequently often listed as threatened by the IUCN (for example, 69% of narrow-ranging bird species are threatened compared to 4% of widespread species; Extended Data Fig. 5). Intermediate and widespread ranging species across taxa are slightly better protected according to representation targets in drylands (amphibians = 18%; birds = 18%; mammals = 16%; reptiles = 16%; Fig. 4); nevertheless, most dryland vertebrate species are inadequately protected.

Fig. 4: Protection of amphibian, bird, mammal and reptile dryland species (≥50% global range in drylands) according to representation targets.
figure 4

Narrow-ranging species = <1,000 km2; intermediate = 1,000–250,000 km2; widespread = >250,000 km2. The protection target for narrow-ranging species is 100% of range size, whereas the protection target for intermediate and widespread species is 10–100% of range size. Scale shows proportion of species with range size protection (IUCN categories I–IV): orange, 0%; blue, 0–10%; green, 10–100%; dark green, 100% range protection. Percentage values above the bar plots are the fraction of each taxon’s protection target covered in protected areas.

Source data

We further explored threat types currently affecting dryland species as designated by the IUCN threat-classification scheme per species (v.3.3; Fig. 5). The largest and most notable anthropogenic threat to vertebrate groups in drylands is agriculture, as reported in other regions45,46,47 (Extended Data Fig. 6). Other prevalent threats in drylands include timber and plant harvesting (for example, harvesting plants and trees for commercial, recreational and subsistence uses), threats from invasive species and disease, and infrastructure development. However, different threats emerge as important across different groups in drylands. For example, amphibians are most impacted by water management in drylands (20% of species assessed) compared to the other groups (Fig. 5a), climate change is a larger threat to birds (which are especially vulnerable to heatwaves)29 in drylands (29% of species assessed; Fig. 5b), over-exploitation (that is, direct hunting and harvesting of animals) is a greater threat to mammals (42% of species assessed; Fig. 5c), and infrastructure development (that is, residential/commercial development and transportation/service corridors) and mining/energy production threaten reptiles relatively more than the other groups (37% and 21% of reptile species assessed, respectively; Fig. 5d). The IUCN has yet to comprehensively review dryland regions and threats to dryland species, and therefore true assessments of threat are not fully known11,48, with potentially important implications regarding our understanding of threats to dryland biodiversity. This is especially true for reptile species—38% in drylands are unassessed or designated as data deficient by the IUCN compared to 24% elsewhere. These knowledge gaps compound the threat to dryland fauna, as the status of unassessed species or those designated as data deficient have been shown to be more similar to species identified as threatened by the IUCN49,50. Overall, most dryland vertebrate species including dryland endemic species and narrow-ranging species are inadequately protected and therefore require more attention in the context of global targets of biodiversity protection, especially due to the vulnerability of drylands to acute land-use threats and to projected future increasing human pressures.

Fig. 5: Threats to terrestrial vertebrate species in drylands.
figure 5

ad, Amphibian (a), bird (b), mammal (c) and reptile (d) threatened species (in IUCN categories CR, EN and VU) in drylands (≥50% global range in drylands). Percentages show the proportion of species assessed affected by threats per taxon. Most species are subjected to multiple threats.

Source data

Current and future land-use change scenarios in drylands

Drylands are especially vulnerable to anthropogenic activities driving biodiversity loss10,13,15,51,52. Furthermore, human populations and rates of urbanization are projected to increase dramatically faster in drylands compared to other regions, especially in Africa and Asia53,54,55. We evaluated cropland, rangeland, pasture and urban land-use pressures in drylands currently and under future climate and global socioeconomic change scenarios (that is, shared socioeconomic pathways, SSPs)56. We calculated the total land cover area of 25 km2 grids occupied by some fraction of the above land types (that is, fractional land-use patterns)57 and additional fractional land areas occupied under future projected scenarios. We find that under different future scenarios by 2100, most drylands across continents are projected to be converted by some degree due to anthropogenic land uses (Fig. 6a–d). At current policy trends, moderate population growth and sustainable consumption of resources and energy (SSP2-4.5), 100% of drylands are projected to be converted to human land uses by some fraction (12.6 × 106 km2 additional natural drylands will be converted; Fig. 6c). This pattern persists even under the most optimistic future scenario (SSP1-2.6), characterized by low emissions with moderate socioeconomic trends limiting global biodiversity loss and environmental impacts (that is, ‘green growth’)57. Again, most natural drylands are projected to be converted by some fraction under such a scenario (9.6 × 106 km2 additional natural drylands will be converted; Fig. 6b), especially as a result of increases in cropland and urban areas with concomitant reductions in pasture and rangeland (including unmanaged grasslands and shrublands with native vegetation)58—a trend we find across different socioeconomic pathways (Extended Data Figs. 710). Therefore, large areas of natural drylands are projected to be converted as a result of increasing cropland and urban areas, while large areas of rangelands (including native shrublands, deserts and so on) are predicted to be converted as well. This is mostly a consequence of the vast land resources needed for agriculture10 and alternative energy production (that is, solar panels and bioenergy)59. Moreover, the development of solar energy resources has wide-ranging impacts on dryland ecosystems and biodiversity through habitat fragmentation and loss60. Also, the substantial expansion of urban land types and infrastructure in drylands including artificial structures and surfaces, roads and mining/extraction sites61,62 are expected to have extensive impacts on dryland species63,64,65 (Fig. 5). Such threats are likely to have multiple and interactive effects on dryland species further exacerbated by climate change and habitat fragmentation66. Similarly, in a projected extreme scenario characterized by rapid and resource-intensive development and high emissions (SSP5-8.5), most natural drylands are projected to undergo some degree of conversion to human land uses (10.4 × 106 km2 additional natural drylands will be converted; Fig. 6d). However, these conversion rates are slightly lower than the ‘middle road scenario’57 (SSP2-4.5; Fig. 6c), mostly due to extreme climate and water restrictions limiting cropland, pasture and rangeland expansion in some dryland regions (Extended Data Fig. 10), coupled with declines in global population67. Ultimately, drylands are projected to face considerable human land-use changes irrespective of projected socioeconomic scenarios. These trends greatly threaten the future integrity of dryland systems with important implications for biodiversity conservation, especially considering that the vast majority of dryland habitats and species are not adequately protected.

Fig. 6: Current and future land-use scenarios in drylands.
figure 6

ad, Current land-use patterns in drylands, LUH2 (2015) (a) and under SSPs: SSP1-2.6 (2100) (b), SSP2-4.5 (2100) (c) and SSP5-8.5 (2100) (d). Colour codes correspond to fraction of grid cell occupied by cropland, pasture, rangeland and urban land classes from low (yellow) to high (blue). Circles show total land cover area of 25 km2 grids occupied by some fraction of land classes above 0% (that is, total fractional land areas). Pink, total fractional land areas in 2015 (a); purple, projected additional total fractional land areas (b–d).

Source data


We evaluated the global distribution of drylands to assess their protected area coverage, unique biodiversity and current and future threats—to highlight the urgent need for the strategic development of conservation targets in drylands. We found that dryland systems host high levels of unique species and habitat diversity that are grossly under-protected. Moreover, natural dryland habitats are perilously threatened by projected human population growth, land-use and climate changes. This is also true under very optimistic scenarios that aim to reduce global agricultural land coverage and promote progressive climate policies, which have dire consequences and other negative spillovers for drylands—for example, the increased conversion of land for alternative energy sources52,68,69. At the same time, drylands provide unique opportunities and considerable scope for achieving conservation and biodiversity goals, especially in Africa and Asia, including the expansion of protected area networks and incorporating stricter forms of protection, while recognizing community-based approaches to conservation70,71 and other management strategies47,72. Greater emphasis on the inclusion of drylands within main frameworks for conservation and development priorities has the potential to notably contribute towards achieving global targets of biodiversity conservation while preserving valuable ecological and human systems in drylands11,14,30.


Global drylands

We evaluated the geographic extent of drylands as classified by the UNEP-WCMC31 using the ‘Drylands Dataset (2007)’. Drylands are categorized based on aridity values defined as the ratio of precipitation to potential evapotranspiration (P/PET), with values below 0.65. Drylands are further distinguished between subtypes using aridity intervals—hyperarid (P/PET < 0.05), arid (P/PET = 0.05–0.20), semiarid (P/PET = 0.20–0.50) and dry subhumid (P/PET = 0.50–0.65). We excluded ‘presumed drylands’—additional areas that do not fall within the above intervals (that is, P/PET ≥ 0.65) but may contain ‘dryland features’—data downloaded from Data were processed using ArcGIS (v10.8.1.), projected using an equal-area Behrmann projection and overlaid with world continents (‘World Continents 2023’; Esri—data downloaded from, omitting Oceania and Antarctica due to their low proportion of dryland area. To determine the proportion of biomes overlapping with drylands (that is, biome area in drylands/global drylands area), we overlaid World Wide Fund for Nature biomes1 with drylands (Extended Data Fig. 1).

Protected area coverage

To assess the degree of protected area coverage of drylands (that is, proportion of protected dryland area), we overlaid drylands with the WDPA32 (data downloaded on November 2023 from Some regions lacked updated data (that is, China), and therefore for China we merged WDPA data from 2016 (when data were last reported fully) with those from 2023. Other regions did not provide complete WDPA data (that is, Turkey and India), and therefore for Turkey and India we compiled data on protected areas using OpenStreetMap33 (data downloaded on November 2023 from; see Supplementary Fig. 1 for comparisons to WDPA 2023 data). We conducted this protected area analysis at two levels: (1) with protected areas defined by the IUCN management categories ‘I–IV’, which are well suited for protecting biological diversity and restricting human activities37 and (2) for all categories of protected areas—using IUCN management categories of protected areas ‘I–IV’ and categories ‘V and VI’ (permitting resource extraction and mixed land uses), including protected areas that lack an explicit IUCN management category for various reasons (that is, categorized as ‘not applicable’, ‘not assigned’ and ‘not reported’)—herein ‘uncategorized’. We omitted protected areas without polygon boundaries. We cleaned the data using the wdpar73 package (v.1.3.7) and the function ‘wdpa_clean’ in R v4.2.2. We excluded ‘Other Effective Conservation Measures’, defined as conservation areas other than a protected area74, because most countries have not yet provided data to the WDPA on Other Effective Conservation Measures47.

Species distribution data, range protection, IUCN threats and representation targets

We obtained the extent of occurrence polygons for all breeding, extant and native amphibians and mammals from the IUCN (v.6.3;, birds from BirdLife International (v.4; and reptiles from the Global Assessment of Reptile Distributions (GARD v.1.7)12. We defined land vertebrate species in drylands as those with ≥50% of their global distributions occurring in drylands (by overlaying species polygons with drylands) for a final dataset of 1,001 amphibians, 2,000 birds, 1,493 mammals and 3,589 reptiles inhabiting drylands. We defined species endemic to drylands as those with >99% of their global distributions occurring in drylands.

For each species, we identified its spatial overlap with protected areas in IUCN categories I–IV (see ‘Protected area coverage’). We categorized species as those with the following: (1) 0% protection (that is, <0.1% range overlap with protected areas), (2) 0–10% protection (that is, ≥0.1% to <10% range overlap with protected areas) and (3) 10–100% protection (that is, ≥10% range overlap with protected areas). We conducted these analyses at the global scale (Fig. 3) and at the continental level (Extended Data Fig. 4 and Supplementary Data 1). We further evaluated the threat status of dryland species, highlighting those species identified as threatened by the IUCN Red List categories—VU, EN and CR—data obtained from For 372 reptile species yet to be assessed by the IUCN (as of June 2022), we used modelled threat assessment categories from ref. 50. We repeated the above analyses for species with ≥90% of their global ranges in drylands and find similar trends as reported above for species with ≥50% of their distributions occurring in drylands (Supplementary Figs 24).

In addition, we evaluated the protection of dryland species by setting representation targets of protection for species according to range size (Fig. 4). We set a 100% protection target for species with ranges <1,000 km2 in drylands and a 10–100% protection target for species with intermediate ranges (1,000–250,000 km2) and widespread ranges (>250,000 km2). We categorized range protection as follows: 0% protection (<0.1% of range size protected), 0–10% (0.1–9.9%), 10–100% protection (9.9–99%) and 100% protection (>99%). We repeated these analyses at the continental level (Supplementary Data 2).

We further assessed the specific types of threat affecting dryland species in IUCN threatened categories (VU, EN, CR; Fig. 5) using the IUCN threat-classification scheme v.3.3—providing comprehensive data on known threats per species (data obtained from We grouped similar threats to allow for simpler comparison according to ref. 45, except for the grouping of ‘invasive species’, which we combined with diseases under the grouping ‘invasion and disease’ according to ref. 46. When relevant, multiple threats were coded per species (for example, ‘agriculture’ and ‘over-exploitation’). Following ref. 46, we considered only future and ongoing threats, we omitted threats affecting only a minority of the global population (that is, <50% of the population) per species, and we removed ‘negligible’ threats and those causing ‘no declines’.

Current and future land-use change scenarios

To investigate current and future human land uses in drylands, we extracted data from the Land Use Harmonization (LUH2) project57 (data obtained from The LUH2 dataset integrates historical human land-use data, management activities and maps with projected models simulating future land-use changes under different scenarios. These models comprise a number of sub-models describing agricultural, demographic, socioeconomic, vegetation and climate systems operating at several different spatial resolutions (for example, local, regional and national)59. Therefore, the harmonization strategy estimates fractional land-use patterns at a globally consistent scale of 0.25° grids integrating historical reconstructions with future projections of land use incorporating multiple sub-datasets. Recently released datasets at higher resolutions—see ref. 75 (for example, their Fig. 10), ref. 76 (for example,their Fig. 4), and ref. 77)—correlate strongly with LUH2 spatially for comparable land classifications; however, these datasets may have several limitations based on data availability yielding potential errors in future land projections. Also, these datasets do not distinguish between grazing areas (that is, rangeland, grassland and pasture regions), which have particular relevance and distinctions in drylands. We compared the LUH2 dataset applied here to drylands with a recently published dataset—the Future Land Use Simulation model76. When comparing global values of total land areas (106 km) of similar land types under projected future scenarios, we find that the general trajectories are similar to LUH2 in drylands (Supplementary Fig. 5). Therefore, we use LUH2—widely applied in spatially explicit global land-use analyses (for example, refs. 78,79,80). However, these data contain several limitations including: their coarse scale, which may cause some deviations in land cover patterns and prevent local analyses; regional differences in estimating future land-use changes based on historical data availability; and additional uncertainties on biodiversity impacts regarding species-specific responses and sensitivities to different land types based on behavioural and physiological traits28,81.

We focused on cropland, rangeland, pasture and urban land-use states, excluding natural vegetation (that is, primary or secondary forest or non-forest), to combine the following raster layers at 25 km2 resolution: ‘managed pasture’, ‘rangeland’, ‘urban land’ and ‘cropland’—comprising ‘C3 annual crops’, ‘C3 perennial crops’, ‘C4 annual crops’, ‘C4 perennial crops’ and ‘C3 nitrogen-fixing crops’. For current conditions, we used LUH2 (v2h) from the year 2015 (providing single land-use estimates per grid without uncertainty intervals). To analyse future trends of land use in drylands, we used Phase 6 of the Coupled Model Intercomparison Project (CMIP6) from LUH2, which provides future scenarios based on alternative climate change and socioeconomic scenarios—SSPs57. We evaluated scenarios SSP1-2.6, SSP2-4.5 and SSP5-8.5. SSP1-2.6 represents a sustainability and low emissions scenario with mean global warming projected at below 1.8 °C by 2100 compared to pre-industrial levels (that is, ‘green growth scenario’). SSP2-4.5 represents a moderate path scenario that does not deviate markedly from current patterns with warming expected at approximately 2.7 °C (that is, ‘middle road’). SSP5-8.5 represents a high-resource and energy-intensive scenario with warming projected to be between 3.3 °C and 5.7 °C (that is, ‘highway’).

We overlaid land cover layers with drylands and identified the land-use fractions of grid cells of exactly 0.22180° resolution (projected using an equal-area Behrmann projection). Grid cells were determined as dryland if grid cell centres fall within the dryland polygon layer. We classified grids according to the following intervals to help with interpretation: (1) 0%, (2) 0–20%, (3) 20–40%, (4) 40–60%, (5) 60–80% and (6) 80–100%. We treated each land class equally without applying different weights54 because impacts are challenging to weigh globally due to non-stationarity and regional variations and sensitivities of different animals and plants to different pressures. To calculate projected future land area of natural drylands converted, we summed the total land cover area of grids occupied by some fraction of the above land classes (that is, all grids above 0% land-use fraction)—herein ‘total fractional land areas’ plus the additional total fractional land areas occupied under future projected SSPs (that is, circles in Fig. 6). Rangelands comprise vast areas of natural and unmanaged habitats and ecosystems including grasslands, shrublands, woodlands and deserts containing native vegetation and typically have low livestocking densities36,58—possibly exerting a disproportionate influence on the land-use patterns reported. Therefore, to quantify the relative contribution of rangeland to total fractional land areas, we repeated the above analysis for cropland, pasture and urban land classes combined excluding rangelands (Extended Data Fig. 7). We also analysed trends in the gain (increase in fraction of each grid cell occupied) or loss (decrease in the fraction of each grid cell occupied) for each land class under different SSPs relative to the baseline (land-use data from LUH2 in 2015) using zonal statistics in ArcGIS (Extended Data Figs. 810).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.