Interactive Atlas

Biocultural Diversity

Where the world's languages and plants are richest — and where they overlap

Two diversities co-vary in space — and disappear together. About 70% of the world's languages live inside biodiversity hotspots that cover roughly 24% of the land (Gorenflo et al., 2012). The places that shelter the most unique plant lineages are, again and again, the same places that shelter the deepest reservoirs of linguistic heritage. This map renders both at the native grain of the source data — no aggregation, no smoothing.

Each of the 8,304 Glottolog points is drawn live as a small luminous splat (so isolated languages stay visible at any zoom and clusters accumulate into density gradients). Plant weighted endemism comes from Daru 2024 (PNAS, Dryad CC0), aggregated from species-level range maps for 201,681 native vascular plants at 20 km grain. Switch to the Richness mode for total species count; climate heterogeneity, the underlying driver, surfaces in the tooltip as context. Hover any point on land to see the overlap.

This interactive visualization requires JavaScript and WebGL2.

Biocultural Diversity languages × plants

Kernel size 7.0 px

About Biocultural Diversity

Two of the great diversities of life on Earth — languages and vascular plants — overlap in striking ways. The most linguistically diverse places on the planet are, again and again, among the most ecologically diverse. This map renders both at once, so the overlap can be seen directly. Climate heterogeneity is the underlying driver, not a third axis: it shapes both the biology and the human geography that the map reveals.

What you're seeing

The default Bivariate view is a Stevens-style bivariate choropleth rendered per-pixel. Two axes are encoded simultaneously by bilinear interpolation between four corner colours, performed in OKLab so the mid-tones are perceptually uniform rather than the muddy averages of additive RGB. The horizontal axis is language richness (the Glottolog kernel-density field). The vertical axis is biological diversity, taken as the geometric mean √(WE × SR) of weighted endemism and species richness — a Pareto composite that only goes high where the flora is simultaneously rich and irreplaceable. Pixels that fall along the diagonal glow gold: those are the biocultural co-hotspots, where the world's deepest reservoirs of language and unique plant lineage sit on the same patch of land.

Why a composite, not endemism alone? Weighted endemism (WE) measures uniqueness — how restricted each species' global range is — and species richness (SR) measures abundance. The Amazon and Congo basin score high on richness but moderately on endemism; Madagascar, the Cape Floristic Region, the Hengduan-Himalaya and Indo-Malay archipelago score high on both. The geometric mean √(WE × SR) is the standard Pareto composite for "irreplaceable diversity": it punishes asymmetry, so a region only enters the top of the axis when richness and endemism are both high. The individual layers remain available via the Endemism and the hidden Richness single-axis modes, but the bivariate view shows the quantity that matters most for conservation.

The four mode toggles at the bottom let you isolate each layer or switch between endemism and richness. Single-axis modes default to the perceptually uniform Turbo colormap (Anton Mikhailov, Google Research, 2019) — purple is low, deep red is the data ceiling. Click the gradient strip in the legend to cycle through Viridis, Plasma, Inferno, Magma, and Cividis as well, all of them perceptually uniform and CVD-aware.

How to use

Hover any luminous dot to see that language's name, family, country, and endangerment status. Hover an empty area for a count of languages within 100 km, the plant α-diversity value, and the climate-driver context for that location.
Mode toggles live at the bottom centre of the map. The legend on the right updates with the current mode.
Biodiversity hotspots can be toggled on from the layers panel — boundaries of the 36 CEPF/Mittermeier hotspots overlay the data in unique colours. Hover any hotspot to highlight its outline and read its name, region, original and remaining extent, and endemic plant and vertebrate counts (Myers et al. 2000 figures where available).
Press P to switch between Equal Earth and Natural Earth projections.

Biodiversity hotspots layer

The optional hotspots overlay shows the boundaries of the 36 biodiversity hotspots as defined by the Critical Ecosystem Partnership Fund and the Mittermeier et al. global hotspot programme — regions that contain at least 1,500 endemic vascular plant species (≥0.5% of the world's total) and have lost ≥70% of their primary native vegetation. Together they cover roughly 2.5% of Earth's land surface but harbour around half of all endemic plants and a third of endemic terrestrial vertebrates.

Each hotspot is drawn in its own hue (golden-angle hue stepping for maximum perceptual separation) so the boundaries stay distinct even where they nest or abut. Toggle the layer from the "Biodiversity hotspots" button at the bottom of the map. The dataset is CEPF Hotspots v2016.1 (Mittermeier et al. 2011, with the Eastern Afromontane and North American Coastal Plain updates), released under CC BY-SA 4.0.

The two diversities (and their driver)

Languages. Roughly 7,000 spoken languages are alive on Earth today. About half are projected to fall silent within this century. The richest concentrations are in places like Papua New Guinea (more languages than all of Europe), the Cameroon highlands, the Caucasus, the Hindu Kush, the Congo basin, and the New World tropics — all of which also rank among the most biodiverse regions on the planet.

Plants. Roughly 340,000 species of vascular plants are alive on Earth today. Total richness (the count of species in a region) peaks in the wet tropics — Amazon, Congo basin, Sundaland — but is a partial signal for biocultural diversity. Weighted endemism goes further and asks how unique a flora is by down-weighting widespread species. The endemism map highlights the Andes, the Cape Floristic Region, Madagascar, the Mediterranean, the Hengduan-Himalaya, and Indo-Malay archipelago — regions whose evolutionary history would be irreplaceable if lost. These align far more closely with linguistic-diversity centres than total-richness alone.

Climate, the driver. The literature on biocultural diversity treats climate heterogeneity as the cause rather than as a third axis of biocultural diversity itself. Sharp gradients in elevation, latitude, and continental position — the Andes, the Himalayas, the East African Rift, the Caucasus — create dense mosaics of micro-climates. Those mosaics produce the niches that hold many plant species, and they isolate the human communities whose languages diverge in parallel. This map surfaces climate in the tooltip as context, not as a co-equal layer, to keep the biocultural framing intact.

Why they go together

The empirical correlation is one of the strongest in human geography. Gorenflo, Romaine, Mittermeier & Walker-Painemilla (PNAS, 2012) showed that the world's 36 biodiversity hotspots and 5 high-biodiversity wilderness areas — covering about 24% of the land surface — contain home ranges for roughly 70% of all languages (~4,800 of ~6,900 at the time of their study). Maffi (Annual Review of Anthropology, 2005) calls this biocultural diversity: a single phenomenon with three faces.

The likely reasons are convergent. Rugged terrain isolates communities; isolated communities diverge linguistically; the same terrain holds many micro-climates, which support many plant species. The places where humans have lived for the longest, in the most varied conditions, are also where the planet's deepest biological and linguistic memory resides.

And they are disappearing in parallel. Sutherland (Nature, 2003) found that languages and species face statistically indistinguishable extinction rates. Garnett et al. (Nature Sustainability, 2018) showed that ~37% of remaining intact ecosystems lie on Indigenous lands — meaning the people whose languages are most at risk are also the custodians of much of what remains of biological diversity.

How it's made

Languages are 8,304 point coordinates from Glottolog 5.3 (Hammarström, Forkel, Haspelmath & Bank 2026, DOI:10.5281/zenodo.18840935, CC BY 4.0) — one per living or recently extinct language, placed at the centroid of its traditional speech area. They are rendered live in WebGL2 as instanced screen-space splats with a dual-Gaussian profile (a tight bright core with a soft halo), additively blended into a half-float buffer so accumulation in clusters preserves the density gradient. Because the splats are in screen space, isolated languages stay visible at any zoom while dense regions form continuous fields.

Plants are from Daru 2024 (PNAS) — predicted native range maps for 201,681 vascular plant species, derived from species distribution models calibrated to dispersal ability and natural habitat, then aggregated to 20 km grid cells. The deposit (Dryad CC0) ships four diversity metrics; we render weighted endemism (WE) by default and offer total species richness (SR) as an alternative. Both are warped from Wagner IV to Equal Earth on the fly, with log1p + z-score normalisation.

Climate context in the tooltip is from Wang et al. 2026 (Conservation Letters) — Rao's quadratic entropy computed over the first four principal components of WorldClim 2.1 in a 5×5 window, bilinear-upsampled to the common grid. It is shown as a hover-only context value, not rendered onto the map.

The plant and climate fields are stored as a single 8,640 × 4,320 RGBA PNG (z-scored channels with a validity mask) and projected to Equal Earth via a CPU-projected lat-lon mesh that the GPU samples with bilinear plus 16× anisotropic filtering. A separate 8,640 × 4,320 water mask is sampled in the fragment shader to discard ocean and inland-water pixels, so the rasters stop crisply at every coastline and lake. The full preprocessing pipeline is open-source at scripts/build-trivariate-sota.py.

Biodiversity hotspots are the CEPF Hotspots v2016.1 polygons (Mittermeier et al. 2011, Noss et al. 2015 for the North American Coastal Plain), simplified with mapshaper Visvalingam and converted to TopoJSON. Released under CC BY-SA 4.0.

References & data

Gorenflo, L. J., Romaine, S., Mittermeier, R. A., & Walker-Painemilla, K. (2012). Co-occurrence of linguistic and biological diversity in biodiversity hotspots and high biodiversity wilderness areas. PNAS, 109, 8032–8037.
Maffi, L. (2005). Linguistic, cultural, and biological diversity. Annual Review of Anthropology, 34, 599–617.
Sutherland, W. (2003). Parallel extinction risk and global distribution of languages and species. Nature, 423, 276–279.
Loh, J., & Harmon, D. (2014). Biocultural diversity: threatened species, endangered languages. WWF Netherlands.
Garnett, S. T., et al. (2018). A spatial overview of the global importance of Indigenous lands for conservation. Nature Sustainability, 1, 369–374.
Hammarström, H., Forkel, R., Haspelmath, M., & Bank, S. (2026). Glottolog 5.3. Leipzig: Max Planck Institute for Evolutionary Anthropology. https://doi.org/10.5281/zenodo.18840935 · CC BY 4.0.
Daru, B. H. (2024). Predicting undetected native vascular plant diversity at a global scale. PNAS, 121(34), e2319989121. https://doi.org/10.1073/pnas.2319989121 · Dataset: Dryad CC0.
Wang, Y., et al. (2026). Mapping the world's climate diversity. Conservation Letters.
Mikhailov, A. (2019). Turbo, an improved rainbow colormap for visualization. Google AI Blog.
Myers, N., Mittermeier, R. A., Mittermeier, C. G., da Fonseca, G. A. B., & Kent, J. (2000). Biodiversity hotspots for conservation priorities. Nature, 403, 853–858. https://doi.org/10.1038/35002501. Source for the per-hotspot endemism counts in the hotspot tooltip.
Mittermeier, R. A., Robles Gil, P., Hoffmann, M., Pilgrim, J., Brooks, T., Mittermeier, C. G., Lamoreux, J., & da Fonseca, G. A. B. (2004). Hotspots Revisited: Earth's Biologically Richest and Most Endangered Terrestrial Ecoregions. CEMEX, Mexico City. Expansion of the Myers 2000 set from 25 to 34 hotspots.
Mittermeier, R. A., Turner, W. R., Larsen, F. W., Brooks, T. M., & Gascon, C. (2011). Global biodiversity conservation: The critical role of hotspots. In Biodiversity Hotspots (pp. 3–22). Springer. CEPF Hotspots v2016.1 · CC BY-SA 4.0.
Noss, R. F., Platt, W. J., Sorrie, B. A., Weakley, A. S., Means, D. B., Costanza, J., & Peet, R. K. (2015). How global biodiversity hotspots may go unrecognized: lessons from the North American Coastal Plain. Diversity and Distributions, 21, 236–244.
Marchese, C. (2015). Biodiversity hotspots: A shortcut for a more complicated concept. Global Ecology and Conservation, 3, 297–309. https://doi.org/10.1016/j.gecco.2014.12.008. Critical synthesis of the hotspot concept and its many parallel definitions.
Amano, T., Sandel, B., Eager, H., Bulteau, E., Svenning, J.-C., Dalsgaard, B., Rahbek, C., Davies, R. G., & Sutherland, W. J. (2014). Global distribution and drivers of language extinction risk. Proceedings of the Royal Society B, 281, 20141574. Companion to the Gorenflo et al. analysis.
Critical Ecosystem Partnership Fund. Biodiversity Hotspots. https://www.cepf.net/our-work/biodiversity-hotspots. Live profile of each of the 36 hotspots.

Loading…

Context & insights

The empirical overlap

Gorenflo and colleagues (2012) showed that the world's biodiversity hotspots and high-biodiversity wilderness areas together cover roughly a quarter of the planet's land surface and harbour about 70% of all human languages — some 4,800 of the ~6,900 spoken at the time. Within the hotspot subset alone, 3,202 indigenous languages are spoken, 1,553 of them by 10,000 people or fewer; many qualify as endangered by population size. The pattern is not statistical accident: linguistic and biological diversity correlate with the same underlying drivers of climate heterogeneity, terrain ruggedness, and historical isolation. Mountains and islands are the planet's twin engines of speciation, both biological and cultural.

What a "hotspot" actually is

Myers and colleagues introduced the modern hotspot concept in 2000, with two strict criteria: a region must contain at least 1,500 endemic vascular plant species (≥0.5% of the world total) and have lost at least 70% of its primary vegetation. Twenty-five hotspots qualified initially; the 2004 revision (Mittermeier et al.) and subsequent work brought the list to thirty-six, the most recent addition being the North American Coastal Plain (Noss et al. 2015), recognised only when fire-suppression-aware mapping showed the longleaf-pine and pitcher-plant ecosystems were as endemism-rich and as fragmented as any tropical hotspot. Marchese (2015) catalogued more than a dozen parallel "hotspot" definitions in active use across conservation science — a useful shortcut, but not a single objective category.

Twin extinctions

Sutherland (2003), comparing IUCN red-list patterns with linguistic-vitality scores, found that languages and species face statistically indistinguishable extinction rates. Amano and colleagues (2014) refined the analysis: the strongest predictor of language-extinction risk is the same as for many vertebrates — small population size in a rapidly changing physical and economic environment. Roughly 25% of the world's languages are spoken by communities of fewer than 1,000 people, and most of those communities live inside the very hotspots highlighted on this map. Loh and Harmon (2014) integrated these signals into a global Index of Biocultural Diversity that tracks the dual loss explicitly.

Why this matters for stewardship

Maffi (2005) framed biocultural diversity not as a metaphor but as a single interlinked phenomenon: the languages, knowledge systems and ecological practices of long-resident peoples are themselves part of the diversity that conservation seeks to protect, and are often the first to vanish when an ecosystem fragments. Garnett and colleagues (2018) estimate that at least 25% of the global land surface is owned, managed or used by Indigenous peoples, and that these lands overlap disproportionately with intact ecosystems and biodiversity hotspots. Conservation frameworks that protect ecosystems without supporting the cultural and linguistic communities that have co-evolved with them tend to fail on both fronts. The hotspot concept, for all its definitional looseness, remains useful precisely because it makes the spatial coincidence visible — and the case for joint stewardship harder to ignore.

References

Primary literature behind the map's data, design and analysis.

Biocultural diversity — concept and analysis

Gorenflo, L. J., Romaine, S., Mittermeier, R. A., & Walker-Painemilla, K. (2012). Co-occurrence of linguistic and biological diversity in biodiversity hotspots and high biodiversity wilderness areas. PNAS, 109(21), 8032–8037. https://doi.org/10.1073/pnas.1117511109.
Maffi, L. (2005). Linguistic, cultural, and biological diversity. Annual Review of Anthropology, 34, 599–617.
Loh, J., & Harmon, D. (2014). Biocultural diversity: threatened species, endangered languages. WWF Netherlands.
Romaine, S., & Gorenflo, L. J. (2014). Spatial congruence in language and species richness but not threat in the world's top linguistic hotspot (Papua New Guinea). Proceedings of the Royal Society B.
Sutherland, W. J. (2003). Parallel extinction risk and global distribution of languages and species. Nature, 423, 276–279.
Amano, T., Sandel, B., Eager, H., Bulteau, E., Svenning, J.-C., Dalsgaard, B., Rahbek, C., Davies, R. G., & Sutherland, W. J. (2014). Global distribution and drivers of language extinction risk. Proceedings of the Royal Society B, 281, 20141574.
Garnett, S. T., et al. (2018). A spatial overview of the global importance of Indigenous lands for conservation. Nature Sustainability, 1, 369–374.

Biodiversity hotspots — definition and revisions

Myers, N., Mittermeier, R. A., Mittermeier, C. G., da Fonseca, G. A. B., & Kent, J. (2000). Biodiversity hotspots for conservation priorities. Nature, 403, 853–858. https://doi.org/10.1038/35002501. Source for the per-hotspot endemism counts in the hotspot tooltip.
Mittermeier, R. A., Robles Gil, P., Hoffmann, M., Pilgrim, J., Brooks, T., Mittermeier, C. G., Lamoreux, J., & da Fonseca, G. A. B. (2004). Hotspots Revisited: Earth's Biologically Richest and Most Endangered Terrestrial Ecoregions. CEMEX, Mexico City.
Mittermeier, R. A., Turner, W. R., Larsen, F. W., Brooks, T. M., & Gascon, C. (2011). Global biodiversity conservation: The critical role of hotspots. In F. E. Zachos & J. C. Habel (Eds.), Biodiversity Hotspots (pp. 3–22). Springer.
Noss, R. F., Platt, W. J., Sorrie, B. A., Weakley, A. S., Means, D. B., Costanza, J., & Peet, R. K. (2015). How global biodiversity hotspots may go unrecognized: lessons from the North American Coastal Plain. Diversity and Distributions, 21, 236–244.
Marchese, C. (2015). Biodiversity hotspots: A shortcut for a more complicated concept. Global Ecology and Conservation, 3, 297–309. https://doi.org/10.1016/j.gecco.2014.12.008.
Critical Ecosystem Partnership Fund. Biodiversity Hotspots. https://www.cepf.net/our-work/biodiversity-hotspots.

Map data sources

Daru, B. H. (2024). Predicting undetected native vascular plant diversity at a global scale. PNAS, 121(34), e2319989121. https://doi.org/10.1073/pnas.2319989121 · Dataset: Dryad CC0. Plant weighted endemism and species richness rasters.
Hammarström, H., Forkel, R., Haspelmath, M., & Bank, S. (2026). Glottolog 5.3. Leipzig: Max Planck Institute for Evolutionary Anthropology. https://doi.org/10.5281/zenodo.18840935 · CC BY 4.0. 8,304 language coordinates.
Wang, Y., et al. (2026). Mapping the world's climate diversity. Conservation Letters. Climate-heterogeneity context layer in the area tooltip.

Cartographic technique

Brewer, C. A. (1994). Color use guidelines for mapping and visualization. In A. M. MacEachren & D. R. F. Taylor (Eds.), Visualization in Modern Cartography. Pergamon. Foundation of bivariate choropleth design.
Stevens, J. (2015). Bivariate Choropleth Maps: A How-to Guide. joshuastevens.net. Modern reference for the four-corner palette construction used here.
Björn, O. (2020). A perceptual color space for image processing — OKLab. Perceptually uniform colour space used for the bilinear interpolation.
Mikhailov, A. (2019). Turbo, an improved rainbow colormap for visualization. Google AI Blog. Default single-axis colormap.