Population Ecology

, Volume 53, Issue 2, pp 373–381

Monitoring tiger populations using intensive search in a capture–recapture framework

Authors

    • Wildlife Institute of India
    • Nature Conservation Foundation
  • Yadvendradev V. Jhala
    • Wildlife Institute of India
Original Article

DOI: 10.1007/s10144-010-0230-9

Cite this article as:
Sharma, R.K. & Jhala, Y.V. Popul Ecol (2011) 53: 373. doi:10.1007/s10144-010-0230-9
  • 190 Views

Abstract

Tigers (Panthera tigris) today face multiple threats to their survival in the form of habitat loss, poaching, depletion of wild prey through illegal hunting and loss of connectivity between populations. Monitoring of tigers is crucial to evaluate their status and react adaptively to management problems. Though camera traps are becoming increasingly popular with researchers enumerating cryptic and elusive animals, they have not been embedded in the regular management activities of tiger reserves. Tiger monitoring, though an important part of the management, is usually implemented using the unreliable pugmark approach. Camera trap-based studies are few, usually of short duration, and are generally conducted by individual scientists and organizations. In this study, we integrate photographic mark–recapture with the routine activity of searching and locating tigers for tourist viewing by the park management in meadows of Kanha Tiger Reserve which form a part of the tourism zone. We validate the density estimates from “tiger search approach” against those obtained from camera trapping and radio-telemetry conducted in conjunction in the same area. Tiger density (\( \hat{D} \) (SE [\( \hat{D} \)]) per 100 km2 for camera traps and tiger search, respectively, was estimated at 12.0 (1.95) and 12.0 (1.76) when effective trapping area was estimated using the half mean maximum distance moved (½ MMDM), 7.6 (1.94) and 7.5 (1.97) using the home range radius, 7.3 (1.49) and 7.5 (1.97) with the full MMDM, and 8.0 (3.0) and 6.88 (2.39) with the spatial likelihood method in Program DENSITY 4.1. Camera trapping, however, was five times more expensive than the tiger search method. Our study suggests that “tiger search approach” can be used as a regular monitoring tool in the tourism zones of tiger reserves, where often most of the source populations are located.

Keywords

Camera trapsMark–recapturePanthera tigrisPopulation monitoringSource populationsSpatially explicit capture–recapture

Introduction

Tigers (Panthera tigris) today face multiple threats to their survival. Apart from habitat loss and degradation and loss of connectivity among populations, poaching for Chinese traditional medicine and the skin trade remain as major threats for long-term survival of tigers. The disappearance of well-known individual tigers from high tiger density tourism zones of a few well-known tiger reserves has caused much concern amongst regular tourists, national media and the general public. Regular monitoring of tigers is crucial to assess population status and to identify problems, so that remedial steps can be initiated in time.

Tiger reserves supporting relatively high densities of tigers are popular tourist destinations, the main motivation for tourists being to see and photograph tigers. Tiger-centric tourism in tiger reserves such as Kanha generates a large amount of revenue, roughly US$ 107,092 per year, and on average a tiger reserve receives 50,000 tourists every year (B. Sinha, personal communication). Tiger sightings in such popular tiger reserves range from 1 to 5 per day. The tourism areas are regularly traversed by the tourist jeeps, naturalists, forest guards and elephant mahouts, in an attempt to locate and see tigers. However, the results of this enormous search effort are not utilized to manage or upkeep a tiger sighting database. At best, the number of tiger sightings in a day is recorded by the park management.

Though capture–recapture-based camera trapping is being increasingly used by researchers for estimating abundance of tigers in South Asia (Karanth 1995; Azlan and Sharma 2003; O’Brien et al. 2003; Kawanishi and Sunquist 2004; Wegge et al. 2004; Simcharoen et al. 2007; Harihar et al. 2009), in most cases it is not being used as a regular monitoring tool. In India, in spite of the falling tiger numbers, tiger monitoring is not done on a regular basis and a countrywide census is done only every 3–5 years, though there are exceptions when scientific monitoring in some tiger reserves has been done on a continuous basis (Karanth et al. 2006), resulting in management solutions and interventions. However such long-term camera trap studies are an exception and still have not found a place in the regular management activities of tiger reserve management. Instead, the pugmark census, which has been criticized in the past as unreliable and unscientific (Karanth et al. 2003), continues to be deployed by park management in the majority of tiger reserves as a regular monitoring tool. Camera trapping being a technology-driven approach is expensive and needs well-trained manpower to implement and execute. Although the popularity of camera traps as a monitoring tool is rapidly increasing, as evident from the increasing number of published studies (Kelly 2008), the camera traps have yet to find a place in the regular management activities of tiger reserves.

Capture–recapture models have seen a gradual development historically (Otis et al. 1978; White et al. 1982), and then tremendous development in the recent past (Pollock et al. 1990; Nichols 1992; Chao and Huggins 2005), resulting in increasingly sophisticated models for estimation of not only population size but also demographic parameters. Since tigers can be individually identified from their unique stripe patterns (Karanth 1995; Franklin et al. 1999), each sighting of a new individual can be treated as a marking event, and capture history can be built from subsequent sightings of new individuals and re-sightings of previously seen individuals.

Effective conservation of tigers should involve consistent monitoring of at least the source populations to ensure that conservation measures put in place are effective and source populations are secure. Thus, it is essential to develop low cost, easy to implement protocols that can be embedded in the routine management activities of a tiger reserve and yet have reasonable precision to detect changes in a population over time. Parts of tiger reserves having relatively high tiger densities are usually designated as tourism zones and are frequented by tourists, domestic as well as international, the major motivation for whom is to see and photograph tigers. In the present study, we integrate tiger photography in a capture–recapture framework to estimate the population size and density of tigers in the tourism zone of Kanha Tiger Reserve, India. We also compared the density estimates from intensive search based capture–recapture with that of camera trapping conducted in the same area (Sharma et al. 2010). Due to a priori knowledge of tigers within the study area, we were reasonably certain that there were 14 individual tigers (9 females, 5 males) and 5 cubs (<1 year) using the study area. A sampling effort of 462 camera trap nights spread over 42 days (14 sampling occasions) had yielded 44 photographs of 12 unique individual tigers (8 females and 4 males). Three cubs were also captured three times; once with the mother and twice without the mother (Sharma et al. 2010). Our objective was to develop and evaluate a quantitative technique that would be cost effective, have sufficient precision to detect changes in the population size, and yet would easily fit in the routine activities of a tiger reserve management.

Materials and methods

Study area

The study was carried out in the tourism zone of Kanha Tiger Reserve in the state of Madhya Pradesh, Central India. Kanha Tiger Reserve is located in the eastern part of the Satpura Range known as the Maikal Hills at 22°45′N, 80°45′E (Schaller 1967) (Fig. 1). Kanha receives an average annual rainfall of around 1,200 mm and the temperature ranges from −3 to 40°C. Kanha has tropical moist deciduous forest dominated by sal (Shorea robusta) and tropical mixed deciduous forest (Champion and Seth 1968). Kanha Tiger Reserve is a typical geo-physiographical representative of the Central Indian Highlands and has been recognized as an important tiger reserve for long-term conservation of tigers (Wikramanayake et al. 1998; Jhala et al. 2008).
https://static-content.springer.com/image/art%3A10.1007%2Fs10144-010-0230-9/MediaObjects/10144_2010_230_Fig1_HTML.gif
Fig. 1

Minimum convex polygon defining the tiger (Panthera tigris) search (53 km2) and camera trap areas (59 km2), buffered with half MMDM. Tiger sighting and camera trap locations are also visible. The mapinset shows location of Kanha Tiger Reserve in the state of Madhya Pradesh, Central India

Major prey species are Chital (Axis axis), Sambar (Cervus unicolor), Barasingha (Cervus duvaucelli branderi), Barking deer (Muntiacus muntjak), Chousingha (Tetracerus quadricornis), Gaur (Bosgaurus), Langur (Semnopithecus entellus), and Wild pig (Sus scrofa). Carnivores included Golden Jackal (Canis aureus), Sloth bear (Ursus ursinus), Wild dog (Cuon alpinus), Leopard (Panthera pardus) and Tiger (Pantheratigris).

Field methods

Tiger search

Kanha meadows, where this study was conducted, forms a part of the tourism zone. The meadows are also the high tiger density zone of Kanha (Karanth and Nichols 1998; Sharma et al. 2010). On average, about 100 four-wheel drive cars traverse every nook and corner of the meadows every day, primarily to see and photograph tigers. Traditionally, tigers in this area are routinely searched for and located by expert local trackers and mahouts on elephant back for tourism and management purposes. Tigers in the tourism zone have thus become habituated to being observed and followed from vehicles and elephants. We used the local knowledge of mahouts, beat guards, tourist photography, camera trapping and knowledge of individual tigers through radio-telemetry to determine the number of tigers a priori (Sharma et al. 2010). An album of photographs of all the individual tigers operating in the study area was developed with a brief description of location–capture history. Each morning, tigers were tracked within the study area with the assistance of 5–8 elephants and experienced mahouts and trackers (Fig. 2). Each elephant commenced its tiger search at 0500 hours and stopped at 0930 hours if a tiger was not located. It covered approximately 5–10 km during this time period. If a tiger was located, it was followed by the elephant until the tiger lay down for the day and its location was then communicated to the Park Management by wireless for tourist viewing. Tourists normally viewed tigers between 0900 and 1130 hours.
https://static-content.springer.com/image/art%3A10.1007%2Fs10144-010-0230-9/MediaObjects/10144_2010_230_Fig2_HTML.jpg
Fig. 2

Tigers being located by an elephant mahout for tourist viewing. A female with two sub-adult cubs can be seen

The elephant-based tracking was complemented by the tiger sighting information from beat guards patrolling their individual beats on foot on a regular basis, and information from naturalists, park guides and tourists traversing the meadows in jeeps. All the three main search methods complemented each other. For example, a beat guard on patrol would visit all the major parts of his beat which are inaccessible to vehicles. Similarly, jeep safaris cover the extensive road network, which is usually not the target area for patrolling by beat guards. The elephant search by experienced mahouts, on the other hand, takes advantage of all the cues of a possible tiger sighting, and their search involves animal trails and even places which do not have trails, as elephants are very well capable of creating a path in the forest. All these three major search methods are used on a daily basis and are independent of each other. The tiger search thus involved a range of sources and a comprehensive search for tigers providing a fairly thorough coverage of the study area. This in turn provides a fairly uniform coverage of the study area and a detailed search within the study area. Thus, it could be safely assumed that no tigers within the study area would have a zero capture probability. We did not, however, have the global positioning system (GPS) track logs of the search area and thus could not quantify the search effort. However, since we were still confident that search effort was comprehensive and did not ignore any parts of the study area, we assumed that the search effort was constant throughout the study area.

Tigers once located were followed until they lay down for the day. Their locations were communicated through wireless and then the research team photographed each of the located tigers from elephant back. An attempt was made to photograph the flanks, face and complete animal from different angles. Every individual tiger was closely observed with 8 × 40 mm binoculars to discern unique markings, injury, special facial patterns, age category, and gender. Every new photograph thus obtained was compared with the album of known individuals to check if it was a new capture or a recapture.

Camera trap survey

We placed cameras at 33 trap sites based on cues such as scent marks, scats, scrapes, and rake marks. Camera placement approached an approximate systematic coverage of the study area, and the average distance between any two camera locations was 1.5 km. The smallest tiger home range in our study area was 12 km2 (Sharma et al. 2010). Based on criteria of placing at least 2–3 camera traps in the smallest home range of the tigers in the study area, there were no major gaps. We used 11 Trailmasters™ (TM 1550) active infrared trail monitoring systems (Goodson Associates, KS, USA) that use an invisible infrared beam across the trail between the transmitter and receiver. For details on the methodology and study design, see Sharma et al. (2010).

Analysis

Individual capture histories for the identified tigers were constructed using a standard X-matrix format (Otis et al. 1978; Nichols 1992) for the tigers photographs captured using camera traps (Fig. 3) as well as the ones photographed during tiger searches. We used models implemented in program CAPTURE (Otis et al. 1978) to derive an estimate of population size (\( \hat{N} \)) and corresponding standard error (SE [\( \hat{N} \)]). Since these models assume that the sampled population does not change in the course of trapping, we tested for violation of closure assumption using the closure test statistic computed by the program CAPTURE for the given capture history data. We also computed the Mo versus Mh, Mo versus Mb and Mo versus Mt test statistic to test for heterogeneity, behavioural response and temporal variation, respectively.
https://static-content.springer.com/image/art%3A10.1007%2Fs10144-010-0230-9/MediaObjects/10144_2010_230_Fig3_HTML.jpg
Fig. 3

A tigress captured by a camera trap in the daytime. Camera trap equipment is visible in the background

The sampled area was estimated as the minimum convex polygon connecting the outermost camera trap locations and outermost locations for sightings of individual tigers. We used the following approaches to estimate the buffer area (\( \hat{W} \)) which was added to the minimum convex polygon to obtain an estimate of effective trapping area (A (\( \hat{W} \))).
  1. 1.

    Half mean maximum distance moved (MMDM) method in which buffer (\( \hat{W} \)) is half of the mean of the maximum distance moved by tigers which are captured more than once (Wilson and Anderson 1985; Karanth and Nichols 1998).

     
  2. 2.

    Full MMDM method in which buffer (\( \hat{W} \)) is MMDM by tigers which are captured more than once (Parmenter et al. 2003).

     
  3. 3.

    Buffer width (\( \hat{W} \)) equal to the home range radius size of the radio-collared tigers is added to the sampled area (Dice 1938, 1941).

     

Tiger search, unlike fixed trapping devices, involves active search for tigers in an area. Since the active search method has not yet been implemented in program DENSITY (Efford et al. 2004), we used a modified approach to enable analysis of tiger search data using maximum likelihood-based spatially explicit capture–recapture (MLSECR) methods. We divided the entire tiger search polygon into 1 × 1 km grids and assigned any tiger seen within the grid to a hypothetical trapping device which was fixed at the center of the grid. This is a viable way of making search data amenable to analysis using MLSECR methods (M. Efford, personal communication). MLSECR methods estimate density directly from the spatial trapping process and do not depend on the addition of a buffer to the sampled area unlike the traditional buffer strip-based methods (Borchers and Efford 2008).

We also computed the cost involved in implementing both the techniques in our study area. For the camera trapping method, cost was computed by taking into account the cost of the camera trap equipment, wages of the research team, operational costs in terms of vehicle hire and fuel, and the cost involved in consumables such as camera films and processing costs. Similarly for the tiger search method, the cost was computed in terms of cost of a long zoom, high resolution digital camera, batteries, wages of the research team, elephant search costs, and vehicle and fuel costs.

The search cost involving elephants was included in this analysis although it is a part of the routine management activities of the Forest Department. This in fact is an income-generating exercise as seeing a tiger from elephant back on an average involves a monetary contribution of 400 INR (US$ 9 approx.) per tourist.

Results

Photographic captures of tigers by camera traps and tiger search

Fifty days of tiger search resulted in 80 tiger sightings of 11 individuals (4 males and 7 females). Three cubs belonging to one litter were seen twice while two cubs belonging to another litter were seen on four occasions.

Of the 11 adult individuals observed during tiger search, 9 were photo-captured by camera traps, while 2 escaped detection by cameras. Of the 2 individuals that were not captured by camera traps, the male was seen twice, while the female was seen four times during tiger search. Both these individuals were seen well within the camera trap grid. Similarly, two females and one male were not recorded during the tiger search, though they were captured through camera traps.

The slope (±SE) of the regression fitted to tiger captures by camera trap data versus sampling effort (1.023 ± 0.025) was smaller in comparison to the slope of the regression fitted to tiger captures by tiger search versus sampling effort (1.72 ± 0.021) (P < 0.01; Fig. 4). Rate of tiger photo-captures through camera traps and tiger search are shown in Fig. 4.
https://static-content.springer.com/image/art%3A10.1007%2Fs10144-010-0230-9/MediaObjects/10144_2010_230_Fig4_HTML.gif
Fig. 4

Rate of tiger photo-captures through camera traps and tiger search

The number of unique individuals (12) stabilized on the 20th day of sampling for camera trap data with 20 photo-captures, while for tiger search data, the unique individuals captured (11) reached saturation on the 27th day of search with 40 sightings (Fig. 5). For camera traps, we obtained a total of 57 captures of tigers which was higher than that of any other animal photo-captured. The percent contribution towards total captures for camera traps and tiger search by the unique individuals captured by both the methods was not significantly different (Fig. 6).
https://static-content.springer.com/image/art%3A10.1007%2Fs10144-010-0230-9/MediaObjects/10144_2010_230_Fig5_HTML.gif
Fig. 5

Cumulative number of unique adult tigers captured through camera traps and tiger search

https://static-content.springer.com/image/art%3A10.1007%2Fs10144-010-0230-9/MediaObjects/10144_2010_230_Fig6_HTML.gif
Fig. 6

Percent contribution of individual tigers towards the overall captures using camera traps and tiger search

Population closure and estimates of population size

The closure test of program CAPTURE failed to reject the closure assumption (z = −1.431, P = 0.076) for the tiger search. Also, the test for behavioural response (Mo vs Mt) did not indicate a behavioural response (χ2 = 1.05, P = 0.30). Testing for heterogeneity (Mo vs Mh) indicated a variation in capture probabilities among the captured individuals (χ2 = 6.39, P = 0.040). However, there was no evidence for temporal variation (Mo vs Mt) in capture probabilities (χ2 = 13.03, P = 1.00).

Using the built-in model selection criteria of program CAPTURE, model Mh (accounting for individual heterogeneity) emerged to be the best fit model (model score = 1.0) for camera trap as well as tiger search data. For tiger search, the model score in decreasing order was 0.97 for Mo, 0.74 for Mtbh, 0.62 for Mbh, 0.50 for Mb, 0.47 for Mth, 0.43 for Mtb, and 0.00 for Mt. We used the jackknife estimator (Otis et al. 1978) available in program CAPTURE to estimate the population size and corresponding standard error. The average capture probability per sample (\( \hat{p} \)) was 0.21 for camera trap data and 0.10 for tiger search data. The population size \( \hat{N} \) with its standard error SE (\( \hat{N} \)) was estimated at 13 (1.76) and 12 (1.63) for camera trap and tiger search data, respectively. The overall probability of capturing a tiger (Mt+1/\( \hat{N} \)) in the sampled area was fairly high (91 and 92%, respectively) for both camera trap and tiger search data.

Estimates of effective trapping area and tiger densities

The minimum convex polygon of 59.2 km2 for camera trap data and 52.7 km2 for tiger search data were buffered with (1) half MMDM (SE) estimated at 1.59 (0.28) for camera trap data and 1.57 (0.33) for tiger search data, (2) full MMDM estimated at 3.19 (0.57) for camera trap data and 3.15 (0.66) for tiger search data, and (3) home range radius estimated at 3.05 (0.80). The estimates of effective trapping area and density estimates including those of MLSECR are given in Table 1. The density estimates by camera traps and intensive search were not different across different methods.
Table 1

Estimates of buffer width (\( \hat{W} \)), number of tigers captured (Mt+1) effectively sampled areas [A (\( \hat{W} \))], number of photo-captures or tiger sightings (n), estimates of population size \( \hat{N} \) (SE (\( \hat{N} \))) using model Mh in program CAPTURE and the corresponding density estimates \( \hat{D} \) (SE (\( \hat{D} \)))

 

Mt+1

(\( \hat{W} \))

A (\( \hat{W} \))

n

\( \hat{N} \) (SE (\( \hat{N} \)))

\( \hat{D} \) (SE (\( \hat{D} \)))

Camera traps

 MMDM

12

1.59

110.72

40

13 (1.76)

12.0 (1.95)

 Full MMDM

12

3.19

178.22

40

13 (1.76)

7.3 (1.49)

 Home range radius

12

3.05

171.65

40

13 (1.76)

7.6 (1.94)

 MLSECR

12

40

8.0 (3.0)

Intensive search

 MMDM

11

1.57

101.00

80

12 (1.63)

12.0 (1.76)

 Full MMDM

11

3.15

164.92

80

12 (1.63)

7.3 (1.19)

 Home range radius

11

3.05

160.40

80

12 (1.63)

7.5 (1.97)

 MLSECR

11

164.37

80

6.88 (2.39)

Comparing operational cost of camera trapping and tiger search

For the duration of the study, the operational cost of carrying out the camera trapping exercise to cover an area of the same size as that covered by tiger search was computed to be US$ 147 per day. The operational cost for tiger search was US$ 34 per day. For camera traps, the operational cost per day, considering the use of 11 camera traps, was broken down into the costs of camera traps: US$ 101; four-wheel-drive jeep: US$ 26; wages: US$ 15 (this included the wage of a trained camera trap person and two assistants); film roll (processing cost not included): US$ 3.34; and a GPS: US$ 2.17. Similarly, the cost of the tiger search was broken down into the cost of a long zoom digital camera: US$ 10.86; a two-stroke bike: US$ 6.52; wages: US$ 8.69 (includes the wages of one trained person); batteries: US$ 1.30. And the cost of sighting a tiger from elephant back: US$ 4.34; and a GPS: US$ 2.17.

Discussion

The rate of tiger captures was higher for tiger search in comparison to camera trap suggesting that tiger search was more efficient at obtaining tiger captures (Fig. 4). However, camera trap data were more efficient in recording new individuals, suggesting a search bias towards more habituated tigers by tiger search (Fig. 5). Both sampling methods reached an asymptote suggesting an adequacy of sampling effort (Fig. 5), but both when considered independently failed to capture all individuals in the population even after prolonged sampling after the asymptote. This suggests that total counts even with intensive sampling are difficult to achieve in natural populations. The rate of tiger captures by camera traps tends to flatten out with sampling intensity suggesting a trap shyness response (Fig. 4). The model selection criteria of the program CAPTURE, however, did not indicate a trap shy response as the behavioral effect model Mb received a very low score (model score = 0.44). However, we minimized trap shyness as traps were moved every day from one block to the other and were camouflaged. Whenever camera avoidance was recorded from track or sign data, we moved the camera location by 50–100 m to another suitable location, and this helped avoid trap avoidance behavior. Though the model selection of program CAPTURE did not suggest a variation in the capture probabilities due to behavioral response or time-specific factors, trap shyness cannot be ruled out. Tigresses rarely take small cubs out on hunting forays (Chundawat 2004), and thus tiger search was better at detecting cubs in comparison to camera traps, as observers could follow a tigress on elephant back to her cubs. The sampling duration for both tiger search and camera trap survey was relatively short (50 and 42 days, respectively) compared to the average lifespan of tigers. Based on the life history traits of tigers, camera trap survey duration of less than 90 days is recommended to avoid the violation of closure (Karanth and Nichols 2002). The percent contribution of individual tigers towards the total captures by camera traps and tiger search was not drastically different, though one particular individual (tiger A) contributed disproportionately towards the total captures in the case of tiger search. This particular individual turned out to be a female which was raising a litter and was largely restricted to a small prey-rich area close to its den, thus making it relatively easy for the elephant mahouts to search for her. Though it cannot be denied that the tiger search method would be positively biased towards dominant and bold individuals, the method did not reduce the probability of capturing shy or less dominant tigers (Fig. 6). Also, tiger search did capture two individual tigers that were missed by camera traps indicating that the method was unbiased towards detecting unique individuals.

The overall probability of capturing a tiger present in the study area (Mt+1/\( \hat{N} \)) was similar for tiger search and camera trap methods and was estimated as 91 and 92%, respectively. The number of photo-captures of tigers exceeded those of any other species. This suggests that, since we had optimized placement of cameras for photographing tigers, photo-captures of other species were not proportionate to their relative abundance. Thus, camera trap studies targeted for a particular species may not provide an index of relative abundance of other species.

We used the heterogeneity model Mh (the best model selected by program CAPTURE) for population estimation both for camera trap and tiger search data. We used this model because the jackknife estimator (Burnham and Overton 1978) for this model is robust to deviations from underlying model assumptions and has performed well in simulation studies (Otis et al. 1978; Burnham and Overton 1979). Also, heterogeneity in capture probabilities is generally expected in natural populations.

The estimates of MMDM by tigers as estimated from camera traps was similar to that obtained by tiger search (see estimates of buffer width W in Table 1). The density estimates obtained from tiger search data matched those of camera trap data across all density estimation models (Table 1).

Tiger search was better at ageing and sexing tigers, as usually we got to spend a considerable amount of time when an individual tiger was sighted. In the long run, this can generate useful data on demographic parameters.

The camera trapping exercise was five times more expensive than the tiger search approach. The cost comparison was made only for the duration of the study which was 50 days. In the long run, if integrated with the management activities of a park, the cost of tiger search would approach near zero as tourists can be requested to deposit their photographs at a central location, while the location can be communicated by the elephant mahouts and park guides, which can then be easily assigned to a specific grid.

The camera trapping exercise involved a consistent and known search effort, as camera traps were placed in a systematic way to provide maximum coverage of the study area and to maximize the capture and recapture probabilities of individual tigers. However, in the case of tiger search, the effort invested may not have been consistent and equal in all parts of the search area. This can result in biases by creating an artificial differential capture probability between tigers occupying more searched and less searched areas. Though theoretically this problem can be addressed by using the heterogeneity model Mh, it is better to avoid this. An inconsistent search effort can lead to complete missing out of some individual tigers, and even heterogeneity models cannot deal well with such scenarios. It is thus advisable to address this problem through a better study design.

In our view, an ideal way of implementing the tiger search approach would be to systematically search the entire sampling area and invest equal search throughout the study area. GPS track logs of the search paths and time spent in the daily search should be maintained to quantify the search effort and mark the outermost boundaries of the search area. This would ensure that the search effort remains consistent and systematic for the entire search area, and that the boundaries of the search area are well defined.

Our study indicates that tiger populations in tourism areas can be estimated and monitored through tiger search in a mark–recapture framework. Since the core of source populations of tigers in India and Nepal are open to tourism (e.g., Kanha, Corbett, Ranthambore, Bandhavgarh, Chitwan, etc.), this method of monitoring these populations provides a relatively inexpensive and less technology-dependent approach that fits into the daily routine of the park management’s objective of locating tigers for tourist viewing. This can be immensely useful for monitoring the core of source populations of tigers, without involving the need to rope in additional resources. By the core of the source population, we imply high tiger density zones within tiger reserves which support much higher tiger densities than the surrounding areas within a tiger reserve. However, one should be cautious while interpreting the results from a monitoring exercise conducted only in a high tiger density zone. Tourism areas are generally relatively safer for tigers as tourism virtually results in a high intensity patrolling of select park areas. Therefore, though a selective approach such as tiger search can serve as an indicator, it would be essential to focus on non-tourism zones as well to assess the overall well being of tiger populations. In 2004, conservationists were shocked by the complete disappearance of tigers from the Sariska Tiger Reserve, a popular tourist destination ironically famous for its tiger sightings. A simple analysis of tiger sightings by tourists (Sankar et al. 2005) would have set the alarm bells ringing, but unfortunately that did not happen.

We are not proposing this approach as an alternative to camera trapping, but as an approach that can fit in the routine management activities of a tiger reserve and yet have sufficient precision to detect changes in the tiger populations or densities. With this approach in place, unfortunate incidents of a tiger population being entirely wiped out unnoticed can be avoided and remedial steps put in place well before it is too late.

Acknowledgments

Rishi would like to thank elephant mahouts, forest guards, park guides and naturalists for their cooperation in detecting and photographing tigers. Our sincere thanks to MD Madhusudan, Abishek Harihar, Kulbushan Suryawanshi, Chandrima Home and Koustubh Sharma for useful suggestions on the earlier versions of this manuscript. We would like to thank Madhya Pradesh Forest Department for permissions for this research and Director, Wildlife Institute of India, for support. We are thankful to the two anonymous referees for their critical comments that greatly helped to improve the quality of the manuscript. We thank Nilanjana Roy and R Raghunath for GIS analysis and Mapping. This project was funded by the Wildlife Institute of India as a part of first author’s Masters Dissertation. Dr. Rajesh Gopal, Director, National Tiger Conservation Authority is thanked for partial funding support for the project.

Copyright information

© The Society of Population Ecology and Springer 2010