# Assessing the viability of tiger subpopulations in a fragmented landscape

> How do you count tigers across 36,000 km² of Sumatran forest? Teams walked trails and ridges looking for tiger tracks and droppings, then camera traps and computer models showed that roads have split the tigers into four groups, three of them small. Keeping those groups connected and free of poaching could decide whether they survive.

Summary by Kersey Lawrence, Original Wisdom. Updated 9 November 2026. Page: https://www.originalwisdom.com/tracking-literature/2006-linkie-chapron-assessing-viability-tiger-subpopulations/

## Citation and access

- **Authors:** Matthew Linkie, Guillaume Chapron, Deborah J. Martyr, Jeremy Holden and Nigel Leader-Williams
- **Year published:** 2006
- **Journal or source:** Journal of Applied Ecology
- **Full citation:** Linkie, M., Chapron, G., Martyr, D. J., Holden, J. and Leader-Williams, N. 2006. Assessing the viability of tiger subpopulations in a fragmented landscape. Journal of Applied Ecology 43(3): 576–586.
- **DOI:** 10.1111/j.1365-2664.2006.01153.x
- **Access status:** Hosted (freely available elsewhere)
- **PDF:** https://www.originalwisdom.com/wp-content/uploads/2026/11/2006_Linkie_and_Chapron_etal_assessing-the-viability-of-tiger-subpopulations-in-a-fragmented-landscape.pdf

## Tags

- **Species:** Mammals > Sumatran tiger
- **Region:** Asia > Indonesia
- **Tracking:** Scat; Sign surveys; Tracking; Tracks; Tracks and signs identification
- **Behaviour and ecology:** Behaviour; Connectivity; Dispersal; Habitat selection; Human disturbance; Wildlife corridors
- **Methods and technology:** Akaike Information Criterion (AIC); Camera traps; Capture-mark-recapture; Detectability; Distribution mapping; Geographic Information System (GIS); Global Positioning System (GPS); Logistic regression; Moran’s I; Occupancy modelling; Population estimation; Simulation
- **Data type:** Modelling or theoretical; New field data
- **Observer qualification:** Not reported
- **Discipline:** Behaviour and ecology; Wildlife tracking
- **Year published:** 2006

## Detailed keywords

- **Species:** Studied: Sumatran tiger (Panthera tigris). Prey: the paper mentions snares set for tiger prey but does not name prey species. Model values for survival were partly based on leopards and cougars. Discussed from other studies: Iberian lynx (Felis pardini), Eurasian lynx (Lynx lynx), puma (Puma concolor), grizzly bear (Ursus arctos) and wolf (Canis lupus), all affected by roads.
- **Location and habitat:** Kerinci Seblat (KS) region, west-central Sumatra, Indonesia (about 36,400 km², including the 13,300 km² Kerinci Seblat National Park). Warm, very humid tropical forest of four types: lowland hill (0 to 300 m above sea level), hill (300 to 800 m), submontane (800 to 1400 m) and montane (above 1400 m). The park has been split in two by farmland, and prey poaching has degraded parts of it.
- **Topics:** occupancy (proportion of area occupied, PAO) and detection probability; true, false and pseudo-absences; sign surveys on trails and ridges; effect of public roads on tiger occurrence; spatially explicit habitat model (SEHM); core habitat areas and subpopulations; tiger density by forest type; population viability analysis (PVA); poaching; habitat connectivity and dispersal of transient tigers; anti-poaching priorities; survey design (fewer cells surveyed more often).
- **Statistics or models:** Occupancy models in PRESENCE software ranked by Akaike Information Criterion (AIC); logistic regression; Moran’s I for spatial autocorrelation; Spearman rank correlation (habitat model against camera trap encounter rates); capture-recapture in CAPTURE software (model Mh, closure tests); encounter rate to density equation of Carbone et al. (2001) with 95% prediction intervals; stage-structured population model in ULM (unified life models) software with demographic stochasticity, 1000 Monte Carlo runs over 50 years. Inconsistency: the text gives core 2 as 5712.6 km², while Table 4 gives a total of 5712.5 km² (its forest-type rows add up to 5712.6).
- **Data type:** New field data: sign surveys in the rainy seasons of 2001 and 2002; camera trapping for 4 to 5 months per site (dates not given); forest map from 2001 Landsat 7 satellite images; logging roads from satellite images 2000 to 2002. Modelling: population viability analysis.

## Synopsis

### 1. What they did

Managers need to know where rare, secretive animals live, how many there are and whether those populations can last, but this is hard to find out quickly for animals that live at low densities over huge areas. The authors set out to answer these questions for tigers in the Kerinci Seblat (KS) region of Sumatra, Indonesia, a top priority area for tiger conservation with little earlier information.

- Sign surveys. They divided the region into 5262 grid cells of 2 × 2 km and chose 200 cells across the main habitats and at different distances from the forest edge. Three teams of field researchers and park staff walked transects (on average 2.47 km long) in each cell, following animal trails and ridges that tigers often use, and recorded whether tiger tracks or droppings were present. Each cell was surveyed twice, the second visit within a week of the first, and 40 cells a third time (440 surveys in total), in the rainy seasons of 2001 and 2002.

- Mapping where tigers occur. Repeat visits let them estimate the chance of finding signs when tigers are there (detection probability). They then tested seven landscape factors (distance to public roads, logging roads, settlements and rivers, altitude, slope and protected status) to see which best explained where tigers were found, and built a habitat map from the best model.

- Checking the map. They compared the map with how often tigers were photographed at 50 camera traps.

- Counting tigers. Camera traps in each forest type (16 to 33 stations, run for 4 to 5 months per site) were used to estimate tiger density, by recognizing individual tigers in photographs (capture-recapture) in three forest types and, in montane forest where only one tiger was photographed, from the photo rate.

- Will they last? They modelled each subpopulation over 50 years with zero to five tigers poached per year, with and without corridors between areas, and with and without full protection from poaching in neighbouring areas.

### 2. What they found

Where tigers occur

- Tiger signs were found in 16.5% of cells on the first visit, 21% on the second and 20% on the third.

- The best model showed that tigers were more likely to be present further from public roads, and that roads also affected the chance of detecting their signs. The detection probability was 0.51.

- Tigers were estimated to use 44.2% of the remaining forest, or 9344 km² of 21,141 km². Good tiger habitat lay mainly inside the national park. Large forest blocks in the north, outside the park, were poor habitat because of nearby roads.

- The habitat map agreed with the camera trap results: more tiger photos came from cells the map rated as better habitat.

How many tigers

- After dropping patches smaller than 200 km² or not connected to the park, four core areas remained.

- Tiger density was 3.3 per 100 km² in lowland hill forest, 2.0 in hill forest, 1.5 in submontane forest and 0.3 in montane forest. The photo rate method gave similar results.

- The four core areas held an estimated 21, 105, 16 and 3 adult tigers (145 in total). Core 2 was the largest (about 5713 km²) and mostly lowland hill and hill forest; the others were mainly higher forest.

Will they last?

- With five tigers poached per year and no connections, the three smaller subpopulations (cores 1, 3 and 4) almost certainly died out (probability above 0.98).

- With three or more tigers poached per year, cores 1 and 3 were not viable, and only the largest subpopulation (core 2) would not die out within 50 years.

- Connections to the large core 2 subpopulation (50% of transients able to move between areas), with poaching stopped there, cut the risk of extinction in cores 1 and 4 to below 1%.

- Protecting a neighbouring area also helped: with two tigers poached per year, core 1 linked to a poaching-free core 3 had a 17% risk of extinction instead of 57%, and core 3 linked to a poaching-free core 1 had a 30% risk instead of 84%.

### 3. What they concluded

- Public roads harm tiger distribution and habitat quality in the KS region and have left four isolated core areas.

- The three larger subpopulations could last if well protected. Anti-poaching effort is best focused on the smaller areas, and keeping core 2 connected to cores 1 and 4 greatly improves the long-term survival of tigers in the park. Cores 1 and 3 are already completely isolated by farmland, so anti-poaching work on core 3 matters most there.

- Tigers still live in tiny core 4, which suggests the dirt road between cores 4 and 2 does not stop them moving.

- Limits: with hindsight, all 200 cells should have been surveyed at least three times. For rare animals it is probably better to survey fewer cells (about 80) more often (three or more times). The population model leaves out genetics and environmental variation for lack of data, and is meant as one part of decision support, not a decision-making tool on its own.

- Yearly sign surveys in each core area could track population trends and show whether conservation is working. The approach could be used for many other secretive animals living at low densities over large areas.

### 4. How tracking was or was not used

Tracking was the main survey method for mapping tigers. Teams walked transects along animal trails and ridges and recorded tiger presence as tracks and/or droppings (faeces). Repeat visits to the same cells allowed the authors to estimate how often signs are missed when tigers are present. No trailing was reported. The paper does not say how tracks or droppings were identified as tiger, how old signs could be, or how many signs were found.

### 5. Tracker or observer certification

Not reported. The sign surveys were done by three teams of field researchers and park staff, with no description of their experience or any test of their skill. This matters because the occupancy and habitat results rest on their sign identification, although repeat visits account for signs they missed.

### 6. Other technologies, skills or innovations

- Camera traps placed on ridges and animal trails, 1 to 4 km apart; individual tigers recognized from photographs.

- Garmin 12XL Global Positioning System (GPS) unit, compass and 1:50,000 maps to record transect locations.

- Geographic Information System (GIS) mapping (ArcView 3.2) with a forest map made from Landsat 7 satellite images.

- PRESENCE software for occupancy models, CAPTURE for capture-recapture, CrimeStat for spatial autocorrelation and ULM for the population model.

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

- Keep the smaller core areas connected to the largest one and keep poaching low in the smaller areas.

- Roads harm tigers by splitting habitat and by giving poachers and farmers access to remote forest; a new road through core 2 would call for more protection there.

- Repeat sign surveys combined with habitat and population models give rapid, practical information for managers, can show stakeholders what is likely to happen without action, and help choose cost-effective strategies.

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

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

- The whole habitat map and population assessment rests on finding and identifying tiger tracks and droppings. Repeat visits correct for signs that were missed, but not for signs that were wrongly identified, so the skill of the survey teams matters, yet it is not reported.

- Sign surveys covered a far larger area than camera traps could, which the authors say would take many years for areas over 10,000 km².

## Cite this summary

Lawrence, K. 2026. Summary of: Assessing the viability of tiger subpopulations in a fragmented landscape (Linkie and Chapron et al., 2006). Original Wisdom, Tracking Literature. https://www.originalwisdom.com/tracking-literature/2006-linkie-chapron-assessing-viability-tiger-subpopulations/