# The shrinking ark: patterns of large mammal extinctions in India

> Where have India’s tigers, lions, rhinos and wild buffalo gone? By matching more than 30 000 old hunting, museum and taxidermy records against what over 100 wildlife experts see today, researchers estimated how often 25 large mammals have vanished from places where they once lived. Protected areas, forest, high ground and people’s tolerance of wildlife all helped some species hold on.

Summary by Kersey Lawrence, Original Wisdom. Updated 9 November 2026. Page: https://www.originalwisdom.com/tracking-literature/2010-karanth-nichols-shrinking-ark-patterns-large/

## Citation and access

- **Authors:** Krithi K. Karanth, James D. Nichols, K. Ullas Karanth, James E. Hines and Norman L. Christensen Jr
- **Year published:** 2010
- **Journal or source:** Proceedings of the Royal Society B: Biological Sciences
- **Full citation:** Karanth, K. K., Nichols, J. D., Karanth, K. U., Hines, J. E. and Christensen, N. L. Jr. 2010. The shrinking ark: patterns of large mammal extinctions in India. Proceedings of the Royal Society B: Biological Sciences 277(1690): 1971–1979.
- **DOI:** 10.1098/rspb.2010.0171
- **Access status:** Hosted (freely available elsewhere)
- **PDF:** https://www.originalwisdom.com/wp-content/uploads/2026/11/2010_Karanth_and_Nichols_etal_the-shrinking-ark-patterns-of-large-mammal-extinctions.pdf

## Tags

- **Species:** Mammals > Asian elephant; Mammals > Asiatic black bear; Mammals > Barasingha; Mammals > Blackbuck; Mammals > Brown bear; Mammals > Chinkara; Mammals > Chital; Mammals > Dhole; Mammals > Four-horned antelope; Mammals > Gaur; Mammals > Golden jackal; Mammals > Grey wolf; Mammals > Indian rhinoceros; Mammals > Leopard; Mammals > Lion; Mammals > Mouse deer; Mammals > Muntjac; Mammals > Nilgai; Mammals > Nilgiri tahr; Mammals > Sambar; Mammals > Sloth bear; Mammals > Striped hyena; Mammals > Tiger; Mammals > Wild buffalo; Mammals > Wild pig
- **Region:** Asia > India
- **Methods and technology:** Akaike Information Criterion (AIC); Distribution mapping; Logistic regression; Occupancy modelling
- **Data type:** Existing data; New field data
- **Observer qualification:** Not needed
- **Discipline:** Population trends
- **Year published:** 2010

## Detailed keywords

- **Species:** Studied (25 large mammals, names as given in the paper): chital (Cervus axis), sambar (Cervus unicolor), muntjac (Muntiacus muntjak), mouse deer (Moschiola meminna), swamp deer (Cervus duvaucelii), blackbuck (Antilope cervicapra), nilgai (Boselaphus tragocamelus), chinkara (Gazella bennetti), four-horned antelope (Tetracerus quadricornis), Nilgiri tahr (Hemitragus hylocrius), wild pig (Sus scrofa), gaur (Bos gaurus), wild buffalo (Bubalus arnee), elephant (Elephas maximus), rhino (Rhinoceros unicornis), black bear (Ursus thibetanus), brown bear (Ursus arctos), sloth bear (Melursus ursinus), hyena (Hyaena hyaena), jackal (Canis aureus), wolf (Canis lupus), wild dog (Cuon alpinus), tiger (Panthera tigris), leopard (Panthera pardus) and lion (Panthera leo).
- **Location and habitat:** All of India, divided into a grid of 1326 cells (average 2818 km², range 2461 to 3059 km²). Covariates included forest cover, elevation, protected areas, human population density and a “cultural tolerance” grouping of Indian states: Rajasthan and Gujarat most tolerant; the seven northeastern hill states, Chhattisgarh and Jharkhand least tolerant; all other states intermediate. Forest, montane, grassland and open forest habitats are discussed.
- **Topics:** local extinction; range contraction; historical distribution records; protected areas; forest cover; elevation; human population density; cultural tolerance of wildlife; imperfect detection; body size; habitat specialists versus generalists; herbivores versus carnivores; crop-raiding and livestock-killing species; endemic species; land-use change; development planning. Inconsistencies in the paper: (1) the Results list mouse deer among the nine species with elapsed time in the top-ranked model, but Table 2 shows no time effect for mouse deer and does show one for sambar; (2) the Results say forest cover had negative coefficients in the top model for three species, but Table 2 shows only two (sloth bear and tiger); sambar’s forest cover coefficient is positive; (3) the Results say the high-elevation prediction was supported for brown bear but not black bear, while Table 2 shows a negative elevation coefficient for black bear and no elevation effect for brown bear; (4) the Results list wild pig and jackal among the species where cultural tolerance lowered extinction, while the Discussion lists them as exceptions; (5) the Methods describe extinction probability as the expected fraction of cells “still occupied”, whereas everywhere else it is the fraction no longer occupied.
- **Statistics or models:** Single-season occupancy models (MacKenzie et al. 2002, 2006) with logit links, fitted by maximum likelihood in program PRESENCE (version 2.0); 50 to 115 candidate models per species; model selection with Akaike’s Information Criterion (AIC) and model averaging of the 2 to 10 top-ranked models; linear logistic models for elapsed time; local extinction estimated as 1 minus occupancy, standardized to 50 and 100 years ago. Covariates: time elapsed since the historic record, protected area presence and proportion (World Database on Protected Areas), forest cover (Global Land Cover 2000), elevation (SRTM digital elevation data), log human population density (LandScan) and cultural intolerance.
- **Data type:** Existing data: more than 30 000 historic records of over 100 mammal species from more than 50 museums, libraries and taxidermy firms, British imperial and district gazetteers, more than 150 hunting journals, the Journal of the Bombay Natural History Society (1885 to 2006) and Indian Forester (1875 to 2006); most from 1860 to 2000, a few from 1760 to 1860. New field data: detection and non-detection reports for 2006 from more than 100 ecologists, naturalists and conservation practitioners (2 to 37 experts per cell), covering the previous year.

## Synopsis

### 1. What they did

Large mammals are among the most threatened animals in the world, and those of South Asia are among the most endangered. The authors asked how much India’s large mammals have disappeared from places where they once lived, and what helps them survive: protected areas, forest, high ground, fewer people, or people who tolerate them.

- Past. They gathered more than 30 000 historic records of mammals from museums, taxidermists, gazetteers, hunting journals and natural history journals, mostly from 1860 to 2000. They chose 25 large mammals with 64 to 3606 reliable records each and mapped them onto a grid of 1326 cells covering India (each about 2800 km²).

- Present. For every cell where a species had once been recorded, they asked more than 100 wildlife experts whether they had detected the species there during the previous year (to 2006). Experts were told that a reported detection had to be certain and based on direct evidence, and that a non-detection would not be read as absence.

- Analysis. They used occupancy models, which allow for animals being present but missed, to estimate the probability that each species had gone locally extinct in each cell. They tested how this was linked to time since the historic record, protected areas, forest cover, elevation, human population density and cultural tolerance, and compared species by body size, diet, habitat and rarity.

### 2. What they found

How much has been lost

- Estimated local extinction over 100 years ranged from 0.14 (jackal) to 0.96 (lion). Other high values were swamp deer (0.90), mouse deer (0.81), Nilgiri tahr (0.71), tiger (0.67), wild buffalo (0.66), wild dog (0.62) and gaur (0.60). Lower values included wild pig (0.25), wolf (0.25), nilgai (0.29), blackbuck (0.33) and sambar (0.34).

- The longer ago a species was recorded in a cell, the more likely it had since disappeared there, for 14 species.

What helped species persist

- Protected areas: linked with lower extinction for 18 species, especially large carnivores and forest-dwelling mammals.

- Forest cover: linked with lower extinction for seven species. For species of open habitats, which have seen intensive land-use change, forest cover alone did not appear to ensure survival.

- Elevation: higher ground was linked with lower extinction for eight species.

- Human population density: more people was linked with higher extinction for 13 species. Wild pig, jackal and blackbuck were unaffected, as predicted; chinkara and nilgai, predicted to be unaffected, were not.

- Cultural tolerance: species did better where people tolerate them, especially nilgai, chinkara and blackbuck in western India, and also several other species.

Which species are most at risk

- Large-bodied species mostly had higher extinction, except elephants, which are culturally tolerated. Smaller species living at naturally low densities (mouse deer, Nilgiri tahr) were also at high risk.

- Habitat generalists (leopard, jackal, wild pig) had lower extinction than habitat specialists (Nilgiri tahr, brown bear, rhino, swamp deer).

- Contrary to prediction, herbivores (0.25 to 0.90) and carnivores (0.14 to 0.96) had similar extinction ranges.

### 3. What they concluded

- All 25 large mammals have had substantial local extinction across India over the past century. Large-bodied, rare and habitat specialist mammals tend to be most at risk.

- Protected areas are critical for reducing local extinction of most of India’s large mammals. Species living mostly outside protected areas (such as blackbuck, nilgai, chinkara, sloth bear, jackal and wolf), and species with only a tiny part of their former range protected (such as tiger, elephant, lion and wild dog), will need new protected areas and connections between existing ones.

- Because the analysis used large grid cells, stronger links may exist at the finer scales where human–wildlife conflict and conservation action happen.

- The results should be considered against rapid changes in land use, climate, human population, culture and economic growth in India and southern Asia.

### 6. Other technologies, skills or innovations

- Combining centuries of museum, taxidermy and hunting records with present-day expert reports to rebuild where species used to live and estimate where they have been lost.

- Occupancy modelling that accounts for imperfect detection, using repeated reports from several experts per cell as replicate surveys.

- Estimates standardized to 50 and 100 years ago, so cells with records from different periods could be compared.

- A “cultural tolerance” covariate grouping Indian states by their historic hunting, diets and reverence for wildlife.

### 7. Implications for conservation (from the paper)

- Many species need new protected areas and better connections between India’s small, fragmented reserves (average under 300 km²) if they are to persist.

- Regional development plans for dams, mines, highways and industry could use these occupancy patterns to reduce harm and avoid further shrinking and fragmenting of mammal ranges.

- Conservation policy needs to take account of changes in land use, climate, human population, culture and economic growth.

### 8. Implications for tracking (Original Wisdom's observations)

_These are Original Wisdom's observations, not the authors'._

- The present-day half of this study rests entirely on more than 100 experts reporting whether they detected each species. The paper asked for certainty based on “direct evidence” but does not say whether this included tracks and signs or only sightings, nor how the experts’ identification skills were checked. Reliable tracks and signs identification, and a way to show it, matters for this kind of country-wide expert survey.

## Cite this summary

Lawrence, K. 2026. Summary of: The shrinking ark: patterns of large mammal extinctions in India (Karanth and Nichols et al., 2010). Original Wisdom, Tracking Literature. https://www.originalwisdom.com/tracking-literature/2010-karanth-nichols-shrinking-ark-patterns-large/