Global impacts of future urban expansion on terrestrial vertebrate diversity

Global urban population is projected to reach 8.5–9.9 billion by 2050 with 55–78% living in urban areas, driving strong demand for urban land. Although urban land covers a small fraction of Earth’s surface, urban expansion is a major driver of land-use change, causing habitat conversion, degradation, fragmentation, and biodiversity loss. Existing assessments indicate substantial biodiversity declines in urbanized areas. With urban area projected to be 1.8–5.9 times 2000 levels by 2100 under various pathways, understanding the rate, magnitude, and spatial distribution of biodiversity impacts from future urban expansion is critical for achieving SDGs (notably SDG 11 and SDG 15) and for informing the post-2020 conservation agenda. Few studies have assessed multidimensional, spatially explicit impacts of urban expansion across scenarios at fine resolution. This study integrates global urban expansion projections under five SSPs with habitat and terrestrial vertebrate biodiversity datasets to quantify future impacts on habitat loss, fragmentation, and losses in species richness and abundance, providing guidance for sustainable urbanization.

Prior global studies show urban expansion contributes to local species richness and abundance declines, but many analyses focus on single impact dimensions (e.g., habitat loss) or use a single scenario and coarse spatial resolution (>1 km). Some forecast urban encroachment on protected areas and biodiversity hotspots, while others assess urbanization’s effect on carbon or specific taxa. There are identified research gaps in quantifying multidimensional biodiversity impacts (habitat loss, fragmentation, species richness and abundance changes) under multiple future socioeconomic scenarios with higher-resolution projections. The study builds on recent global urban expansion projections (SSP-consistent) and biodiversity response models (PREDICTS) to address these gaps.

Urban expansion scenarios: Used global, spatially explicit projections of urban expansion (1 km, decadal 2020–2100) under five SSPs (SSP1–SSP5) from Chen et al. (2020). Urban land demand per capita was estimated via panel regressions using historical GHSL urban land and socioeconomic data (urbanization rate, per capita GDP), then multiplied by projected population to get regional demand for 32 macro SSP regions. Spatial allocation employed the FLUS model to simulate urban growth patterns, coupling human and environmental drivers. Assumptions included: urban land conversion is irreversible; existing urban area remains if future demand is lower; urban land demand can still grow with population decline if urbanization and GDP increase. Habitat loss assessment: The 2015 ESA CCI Land Cover was reclassified into cropland, urban land, forest, shrubland, grassland, and others; all except cropland and urban land were treated as natural habitats. Spatial overlap between projected urban expansion and 2015 natural habitats quantified direct habitat loss globally, by biome, and by ecoregion. Optimized Hot Spot Analysis (ArcGIS Pro) identified statistically significant hot/cold spots (Gi_Bin) of habitat loss. Protected areas (WDPA) were intersected with projected urban land to estimate urban encroachment assuming no strict constraints within PAs, based on observed 1992–2015 trends. Impacts on biodiversity prioritization schemes were quantified for Biodiversity Hotspots, WWF Global 200 ecoregions, and Last of the Wild areas. Habitat fragmentation: Proximity effects were measured as mean Euclidean distance from urban patch edges to nearest natural habitat patch edges at 1 km resolution for 2015 and each future time step; inter-annual changes indicated increasing adjacency. Additional fragmentation metrics—mean patch size (MPS), edge density (ED), and mean Euclidean nearest-neighbor distance between habitat patches (ENN_MN)—were computed with FRAGSTATS at the ecoregional scale within a 5 km buffer of urban land in 2100 for each SSP. Land-systems and biodiversity response: An updated global land-systems map (1 km) integrating land cover, use intensity, and livestock density was refined by replacing urban land with 2015 ESA CCI urban data and reconstructing classes where needed via decision-tree rules. Species richness data (10 km equal-area grid) for amphibians, mammals, and birds (including threatened and small-ranged subsets) were from Jenkins et al. The PREDICTS database provided model-based percentage changes in local species richness and total abundance relative to a natural baseline across land-use intensities. Estimating biodiversity loss: For each 1 km grid cell projected to convert from a non-urban land system to urban, relative percentage changes in local species richness and total abundance were estimated by mapping land-system classes to PREDICTS land-use intensity classes and computing the difference in biodiversity between original land system and urban. For absolute species richness losses, within each 10 km cell the area-weighted mean percentage loss due to urban conversion (2020–2100) was multiplied by the baseline number of species to estimate potential numbers of species lost. Threatened and small-ranged species losses were mapped using their distribution layers. Scenario-specific and average impacts were summarized at global, regional, national, and ecoregional scales.

- Urban expansion scale: By 2100, 36–74 Mha of land will become urban (54–111% increase over 2015), causing 11–33 Mha of natural habitat loss across SSPs. - Spatial patterns of habitat loss: Hotspots include US coasts (NE, South, West), Gulf of Guinea, Sub-Saharan Africa, and Persian Gulf coasts; under SSP5, central and western Europe become hotspots. East and South Asia (China, India, Japan) are predominantly cold spots after mid-century due to declining demand. - Biomes and ecoregions: Largest total natural habitat loss in temperate broadleaf and mixed forests (except SSP3). Temperate broadleaf and mixed forests may decrease by 1.4% under SSP5. About 9% of 867 terrestrial ecoregions will lose >1% habitat; four ecoregions exceed 20% loss: Atlantic coastal pine barrens, coastal forests of the northeastern US, Puerto Rican moist forests, and Puerto Rican dry forests. - Protected areas (PAs): In 2015, 30,594 km² of urban land occurred within 28,152 PAs (12.6% of PAs), with 38% of PA urban land change (1992–2015) from natural habitat conversion. Without strict restrictions, by 2100, 13.2–19.8% of PAs will contain urban land, spanning 29,563–44,400 PAs and 46,705–89,901 km² of urban area. - Biodiversity hotspots: Additional 1.5–1.8% of hotspot area will be urbanized by 2100. Hotspots with highest proportional urban conversion: California Floristic Province (6–11%), Japan (6–8%), North American Coastal Plain (4–8%), Guinean Forests of West Africa (4–8%), Forests of East Australia (2–6%). Some hotspots with low 2015 disturbance will see disproportionate increases (e.g., Guinean Forests of West Africa +281–708%; Coastal Forests of Eastern Africa +294–535%; Eastern Afromontane +169–305%; Polynesia–Micronesia +33–337%). East Melanesian Islands and New Caledonia are minimally affected. - WWF Global 200: ~93% of Global 200 ecoregions will be affected. Although per-ecoregion urban share remains <1%, urban area within them increases 74–160% (2015–2100). Highest urban growth rates: Sudd–Sahelian Flooded Grasslands and Savannas (+877–9955%), East African Acacia Savannas (+527–646%), Hawaii Moist Forest (+18–902%), Congolian Coastal Forests (+500–1037%). - Last of the Wild: Urban area increases by only 73–213 km² by 2100. - Proximity and fragmentation: Urban areas will move closer to edges of 34–40 Mha of natural habitat by 2100; typical edge-to-edge distance reductions: ~2000 m (wetland/other/forest), ~1500 m (grassland), ~900 m (shrubland). Countries with large proximity changes include Mauritania, Algeria, Saudi Arabia, Western Sahara, and the US. Within 5 km of urban areas, MPS declines (most preserved under SSP1, most fragmented under SSP5), ED increases (lowest under SSP1; rapid late-century rise in SSP5), and ENN_MN increases, indicating greater isolation; SSP1 best maintains proximity/connectivity. - Local biodiversity responses (1 km): Average relative loss in species richness due to conversion to urban land is 34%; average relative loss in total abundance is 52%. Highest risks for conversions from dense forest, mosaic grassland/open forest, mosaic grassland, and bare/natural grassland to urban (e.g., richness loss 48%, 95% CI 34–59%; abundance loss 62%, 95% CI 38–76%). - Absolute species losses (10 km): Average loss of 7–9 species per cell by 2100 across scenarios; maximum losses up to ~197 species per cell. Losses concentrate in Sub-Saharan Africa (Gulf of Guinea), the US, and Europe; also elevated in SE Brazil, India, and eastern Australia. Under SSP5, species richness loss occurs over ~740 Mha. Under SSP3–SSP4, richest hotspots in Sub-Saharan Africa and Latin America are disproportionately threatened. - Threatened and small-ranged species: Up to 12 threatened species and up to 40 small-ranged species lost per 10 km cell in worst-affected areas. Hotspots for threatened losses include West/East Africa, northern India, and eastern Australia; small-ranged losses are more spatially concentrated. - Priority areas and protection gaps: 30 conservation-priority ecoregions (all in Latin America and Sub-Saharan Africa) identified for combined high habitat loss and small-ranged species loss risk. Only 6.4–8.6% of the top 5% highest-loss 10 km cells are covered by current PAs. Countries with the largest potential species losses (>29 species per 10 km cell) include Kenya, Eswatini (Swaziland), Brunei, Zambia, Republic of Congo, and Zimbabwe. - Development pathway effects: SSP1 (sustainability) consistently yields the lowest habitat loss, fragmentation, and biodiversity loss; SSP5 tends toward larger late-century impacts.

Future urban expansion will significantly affect terrestrial biodiversity through direct habitat loss, encroachment on biodiversity priority areas, increased fragmentation, and reductions in species richness and abundance. The magnitude and spatial distribution of impacts depend strongly on socioeconomic pathways. Sustainable development (SSP1) minimizes habitat loss and fragmentation and better conserves species, while fossil-fueled/high-consumption development (SSP5) exacerbates late-century impacts. Impacts are spatially heterogeneous across habitats, hotspots, biomes, and ecoregions; areas with currently low human disturbance in biodiversity-rich regions (often in lower-income countries) are projected to face disproportionately high urban growth and biodiversity risk, underscoring needs for proactive conservation planning and funding. Urban expansion disproportionately threatens habitats adjacent to cities, increasing edge effects and isolation; compact urban forms and ecological corridors can help maintain connectivity. Despite the smaller area of urban expansion relative to cropland, per-area biodiversity losses from urban conversion are often largest due to conversion to built environments. Concentrating conservation in a limited number of high-risk ecoregions could avert disproportionate species losses, but many of these areas are underrepresented in current protected networks and conservation spending. Integrating urbanization considerations into biodiversity frameworks and coordinating SDGs are essential, with emphasis on nature-based solutions, sustainable planning, and cross-border governance for urban agglomerations.

This study integrates high-resolution, scenario-based global urban expansion projections with habitat and terrestrial vertebrate datasets to quantify future impacts on habitat loss, fragmentation, and species richness and abundance. It shows that by 2100, urban expansion will cause substantial natural habitat loss (11–33 Mha), intensify fragmentation near cities, and reduce local biodiversity (average −34% species richness, −52% abundance), with hotspot impacts in parts of Africa, the Americas, Europe, and Australia. Biodiversity prioritization areas, including hotspots and WWF Global 200 ecoregions, will be disproportionately affected, and current protected area networks cover only a small fraction of the most impacted cells. A sustainable urban pathway (SSP1), compact development, ecological corridors, and improved, outcome-focused protected area categorization can mitigate impacts. Future research should incorporate indirect, telecoupled land-use effects, dynamic global land cover and biodiversity data, finer-scale urban demand projections beyond macro regions, realistic constraints within protected areas, and interactions with climate change to better forecast and manage biodiversity outcomes under urbanization.

- Indirect effects not modeled: Telecoupled impacts (e.g., trade-driven land-use displacement) stemming from urban expansion were not quantified and may shift biodiversity loss to distant locations. - Data constraints: Lack of long-term, high-resolution global simulations of dynamic land cover/land-system change and dynamic biodiversity limited temporal precision of loss estimates and precluded assessing biodiversity responses to fragmentation directly. - Scenario/regional aggregation: Urban demand projections are at 32 macro-SSP regions; this aggregation introduces uncertainty for national and subnational spatial allocations and impact estimates. - Protected area assumption: The analysis assumes continued potential for urban expansion within protected areas (reflecting historical trends), potentially overestimating direct encroachment in places with strong enforcement; conversely, urban growth would be displaced elsewhere if constrained, affecting other habitats. - Climate interactions omitted: Feedbacks between urban expansion and climate change (e.g., urban heat islands, extreme events, sea-level rise) and climate-driven species redistribution were not included but could alter future biodiversity patterns and impacts.