# Tigers and their prey: Predicting carnivore densities from prey abundance

> How many tigers can a forest hold? Across 11 reserves in India, researchers walked 6,820 km counting deer, pigs and wild cattle, and set camera traps that recognised 167 tigers by their stripes. Tiger numbers closely followed a simple prediction of about one tiger for every 500 hoofed prey animals, pointing to prey loss as a main driver of tiger declines.

Summary by Kersey Lawrence, Original Wisdom. Updated 9 November 2026. Page: https://www.originalwisdom.com/tracking-literature/2004-karanth-nichols-tigers-their-prey-predicting/

## Citation and access

- **Authors:** K. Ullas Karanth, James D. Nichols, N. Samba Kumar, William A. Link and James E. Hines
- **Year published:** 2004
- **Journal or source:** Proceedings of the National Academy of Sciences of the United States of America (PNAS)
- **Full citation:** Karanth, K. U., Nichols, J. D., Kumar, N. S., Link, W. A. and Hines, J. E. 2004. Tigers and their prey: predicting carnivore densities from prey abundance. Proceedings of the National Academy of Sciences of the United States of America 101(14): 4854–4858.
- **DOI:** 10.1073/pnas.0306210101
- **Access status:** Hosted (freely available elsewhere)
- **PDF:** https://www.originalwisdom.com/wp-content/uploads/2026/11/2004_Karanth_and_Nichols_etal_tigers-and-their-prey-predicting-carnivore-densities-from-prey-abundance.pdf

## Tags

- **Species:** Mammals > Barasingha; Mammals > Chinkara; Mammals > Chital; Mammals > Four-horned antelope; Mammals > Gaur; Mammals > Hog deer; Mammals > Muntjac; Mammals > Nilgai; Mammals > Sambar; Mammals > Tiger; Mammals > Wild buffalo; Mammals > Wild pig
- **Region:** Asia > India
- **Behaviour and ecology:** Behaviour; Predation
- **Methods and technology:** Bayesian statistics; Camera traps; Capture-mark-recapture; Detectability; Distance sampling; Markov chain Monte Carlo (MCMC); Population estimation
- **Data type:** Modelling or theoretical; New field data
- **Observer qualification:** Not needed
- **Discipline:** Behaviour and ecology
- **Year published:** 2004

## Detailed keywords

- **Species:** Studied: tiger (Panthera tigris) and their principal wild ungulate prey: wild pig (Sus scrofa, 11 sites), sambar (Cervus unicolor, 10 sites), axis deer (Axis axis, 9 sites), gaur (Bos gaurus, 7 sites), muntjac (Muntiacus muntjak, 6 sites), four-horned antelope (Tetracerus quadricornis, 6 sites), nilgai (Boselaphus tragocamelus, 5 sites), barasingha (Cervus duvaceli, 2 sites), chinkara (Gazella bennetti, 2 sites), wild buffalo (Bubalus bubalis, 1 site) and hog deer (Axis porcinus, 1 site). Other predators mentioned but not studied: leopard (Panthera pardus) and dhole (Cuon alpinus). Livestock mentioned as possible alternative prey at Melghat.
- **Location and habitat:** 11 protected areas across India, 3,024 km² in total: Melghat, Tadoba, Pench-MR, Bhadra, Panna, Bandipur, Nagarahole, Kanha, Kaziranga, Ranthambore and Pench-MP. Dry forest, moist forest, mixed moist and dry forest, and alluvial grassland (Kaziranga), with annual rainfall from 800 mm (Ranthambore) to 3,000 mm (Kaziranga).
- **Topics:** predator–prey relationships; prey as a limiting factor for large carnivores; predicting tiger density from prey density; kill rates (about 50 ungulates per tiger per year); macroecology; indices versus estimates that account for detection; detection probability; individual identification from stripe patterns; prey depletion by poaching, hunting, livestock grazing and villages inside reserves; tiger conservation.
- **Statistics or models:** A priori mechanistic model (tiger density = 0.10/50 × prey density); line transect distance sampling (program DISTANCE; half-normal, hazard rate and uniform–cosine key functions); photographic capture–recapture for closed populations (program CAPTURE, model Mh with the jackknife estimator); effectively sampled area from the trap polygon plus a buffer of half the mean maximum distance moved; hierarchical Bayesian regression of log tiger density on log prey density (bivariate normal), fitted with Markov chain Monte Carlo (MCMC) in WinBUGS; a reduced model with the slope fixed at 1. Small inconsistency: Table 1’s predicted tiger densities equal 0.2 × prey density for nine sites, but Melghat (5.3 prey per km², predicted 1.04 rather than 1.06) and Tadoba (13.1, predicted 2.61 rather than 2.62) differ slightly, probably from rounding of the prey estimates.
- **Data type:** New field data from a long-term study, 1995–2003 (dates for individual sites not given): 6,820 km of line transects and 8,677 camera trap-days. Modelling: an a priori model tested against the field estimates.

## Synopsis

### 1. What they did

Ecologists would like to predict how many animals of a species an area can hold from the few resources that limit them. In practice this has proved very hard, partly because large-scale studies often rely on raw counts (indices) that ignore how many animals were missed. The authors set out to do it properly for tigers.

- A simple prediction. Tigers eat almost only large hoofed animals (ungulates), and earlier studies suggested that predators remove about 15% of the available prey each year, with tigers taking about 10% and leopards and dholes the rest. With each tiger killing about 50 ungulates a year, the model predicts roughly one tiger for every 500 ungulates (0.2 tigers per 100 km² for each ungulate per km²).

- Counting prey. At 11 protected areas across India, teams of trained investigators walked forest trails as line transects, counting groups of ungulates and measuring how far they were from the line. Distance sampling then estimated how many animals were missed and gave prey density.

- Counting tigers. Automatic camera traps photographed tigers, which were identified individually from their stripes. Capture–recapture analysis of these photographs estimated how many tigers lived at each site and their density.

- Testing the prediction. They compared tiger and prey densities with a Bayesian model, using 9 of the 11 sites. Two sites were left out for reasons stated in advance: at Pench-MP, heavy poaching just before the survey was suspected to have pushed tigers below what the prey could support; at Kaziranga, other large predators were virtually absent, so tigers were expected to be denser than elsewhere.

### 2. What they found

Prey

- The teams walked 6,820 km and detected 8,061 groups of prey.

- The chance of seeing an animal within the surveyed strip ranged from 0.2 to 0.8 across species and sites.

- Wild ungulate density ranged more than tenfold, from 5.3 per km² at Melghat to 63.8 per km² at Pench-MP.

- Well-protected sites (Kanha, Nagarahole, Pench-MP, Ranthambore, Kaziranga) held 56.1 to 63.8 ungulates per km². Panna and Bandipur were lower (30.9 and 35.2), which the authors put down to less effective protection. Bhadra, Tadoba, Pench-MR and Melghat (5.3 to 16.8) appeared well below their potential because of villages inside the reserves.

Tigers

- 8,677 camera trap-days photographed 167 individual tigers, from 5 at Pench-MP to 26 at Kanha.

- The cameras photographed only 38% to 86% of the tigers estimated to be present, which shows why detection has to be accounted for.

- Tiger density ranged from 3.27 per 100 km² at Tadoba to 16.76 per 100 km² at Kaziranga.

Tigers and prey together

- With no constraint on the shape of the relationship, the slope was estimated at about 0.5, suggesting tiger density may rise less steeply than prey density, but the evidence against the predicted slope of 1 was not conclusive.

- With the slope fixed at 1, the estimated ratio was 0.247 (95% credible interval 0.181 to 0.336), entirely consistent with the predicted 0.2.

- Pench-MP and Melghat fell outside the prediction range. Melghat had more tigers than predicted, which the authors speculate may be partly due to large numbers of livestock as alternative prey.

### 3. What they concluded

- The model predicted tiger densities from prey abundance well, giving evidence of a real link between the numbers of large carnivores and their prey across a wide range of conditions. Some uncertainty remains about the exact form of the relationship.

- Two things set the study apart: predictions made in advance from a simple model of what limits tigers, and density estimates that account for animals missed during counts rather than relying on indices.

- Ecological studies can be done rigorously at landscape scale, even though it took 8 years and a large field effort.

### 6. Other technologies, skills or innovations

- Automatic camera traps triggered by movement, with individual tigers identified from the shape and arrangement of stripes on their flanks, limbs and faces.

- Line transect distance sampling for prey, analysed in program DISTANCE.

- Program CAPTURE for closed-population capture–recapture, and WinBUGS for the Bayesian analysis (code and output were made available online by the U.S. Geological Survey).

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

- Prey density is a key determinant of how many large cats an area can hold.

- At Bhadra, Tadoba, Melghat and Pench-MR, tiger densities appear to have fallen as people depleted the prey, even though these sites are ecologically similar to high-density sites.

- The results are consistent with the idea that wild tiger declines are mainly caused by prey depletion driven by human impacts, so reducing these impacts through management should be a central concern of conservationists.

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

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

- The study did not use tracking, but its central point applies to track-based surveys too: a raw count (of animals, or of tracks) only reflects how many animals are present if the proportion detected stays the same across places, times and species. Here, the proportion seen ranged from 0.2 to 0.8 for prey and 0.38 to 0.86 for tigers.

## Cite this summary

Lawrence, K. 2026. Summary of: Tigers and their prey: Predicting carnivore densities from prey abundance (Karanth and Nichols et al., 2004). Original Wisdom, Tracking Literature. https://www.originalwisdom.com/tracking-literature/2004-karanth-nichols-tigers-their-prey-predicting/