OUP user menu

Temporal Variation in Tiger (Panthera tigris) Populations and its Implications for Monitoring

Adam C. D. Barlow , Charles McDougal , James L. D. Smith , Bhim Gurung , Shiv R. Bhatta , Sukram Kumal , Baburam Mahato , Dhan B. Tamang
DOI: http://dx.doi.org/10.1644/07-MAMM-A-415.1 472-478 First published online: 14 April 2009

Abstract

Tigers (Panthera tigris) are endangered wild felids whose elusive nature and naturally low densities make them notoriously difficult to count. We present 7 years of camera trapping, tracking, and observational data on a local tiger population in Chitwan National Park, Nepal, to quantify changes in abundance of demographic groups and to identify underlying causes. Mean abundance in the 100-km2 study area was 18 tigers, but there was high variance among years. Transients were generally recorded at low levels, but there were large oscillations in numbers of tiger offspring, driven by infanticide perpetrated by male tigers taking over territories. The number of breeding animals in the study area remained relatively stable, with about 6 breeding females and 1 or 2 breeding males. The high density of breeding adults in Chitwan National Park highlights the region as a potential stronghold for tigers. Concentrating on counting breeding animals increases the power of monitoring programs to detect change over time. An alternative approach is to carry out surveys on a scale large enough to encompass sufficient territories to compensate for the local impacts of periodic turnover of adult males on total abundance.

Key words
  • camera trapping
  • infanticide
  • Nepal
  • Panthera tigris
  • power analysis
  • tiger
  • tracking

Wild tigers (Panthera tigris) are threatened throughout their range by habitat degradation, poaching, and depletion of natural prey (Carroll and Miquelle 2006; Dinerstein et al. 2007; Sanderson et al. 2006). Management strategies to secure the last remaining tiger populations can be evaluated by assessing changes in tiger numbers: increasing or stabilized numbers indicate success, whereas decreasing populations may help identify threats and highlight the need for intervention. Choosing when to react to a perceived change in tiger numbers will be a site-specific judgment based on management objectives and risk assessment (Mapstone 1995). However, any detected change may be a product of anthropogenic factors and natural temporal fluctuations in the population (Gibbs et al. 1998). Appropriate monitoring strategies are necessary to take into account the scale and range of these fluctuations in order to make informed management decisions (Gibbs et al. 1998).

Camera trapping, radiotelemetry, and secondary sign surveys all have been used to estimate tiger abundances at various sites, but most studies have been limited to making “snapshot” assessments, rather than tracking change over time (Johnsingh and Negi 2003; Johnson et al. 2006; Karanth 1995; Karanth and Nichols 1998; Karanth et al. 2004a; Kawanishi and Sunquist 2004; Linkie et al. 2006; Lynam et al. 2007; O'Brien et al. 2003; Rabinowitz 1993; Simcharoen et al. 2007; Sunquist 1981). Ongoing work in Russia, India, and Bangladesh (Barlow et al. 2008; Hayward et al. 2002; Karanth et al. 2006) is measuring overall trends in tiger populations, but only Smirnov and Miquelle (1999) have done this with respect to different demographic groups. However, their study was conducted on a recovering population that was far below carrying capacity at the start of their 28-year study. No publication has to date documented the natural temporal fluctuations of demographic groups for a population close to carrying capacity. Distinguishing between demographic groups is important, because how particular sectors of the population contribute to long-term viability and the degree to which they fluctuate may influence management strategy and monitoring design.

The tigers in Chitwan National Park, Nepal, have been studied continuously since the 1970s (McDougal 1977; Seidensticker 1976; Smith 1993; Smith et al. 1987, 1998a, 1998b; Smith and McDougal 1991; Sunquist 1981). As an extension of this long-term program, we present 7 years of camera-trap, secondary sign, and observational data to quantify the temporal fluctuations of different demographic groups in a local tiger population close to carrying capacity, identify underlying causes of that change, and evaluate how the scale of temporal fluctuations in tiger abundance may affect the power of monitoring programs to detect change over time.

Materials and Methods

Study area.—Located in south-central Nepal, Chitwan National Park (27.34–27.69°N, 83.87–84.75°E) was granted official national park status in 1973 with a designated area of 544 km2. Three years later, the park was expanded to its current size of 932 km2. Elevations in the park range from 134 to 914 m above sea level and the park receives an average annual rainfall of 2,400 mm. The vegetation is dominated by moist deciduous forest, riverine forest, and grassland (Mishra 1982). The park is part of the larger Chitwan-Parsa-Valmiki Tiger Conservation Unit that encompasses 2,543 km2 of contiguous tiger habitat (Smith et al. 1998a; Wikramanayake et al. 1998). Chitwan National Park also is part of the larger Terai Arc Landscape conservation area, which has been established to help secure the tiger metapopulation in the region (Wikramanayake et al. 2004).

We monitored changes in tiger abundance in a 100-km2 patch of alluvial floodplain habitat in a western sector of the park. The northern and southern boundaries of this area were formed by the Rapti River, which marked the boundary of the park, and the Churia Hills, respectively. The eastern and western boundaries were set using a priori information on the territorial spacing of some resident female tigers, deduced by more than 20 years of previous monitoring using secondary sign surveys and telemetry (McDougal 1977; Smith 1993; Smith and McDougal 1991; Smith et al. 1987, 1998b; Sunquist 1981).

Measuring temporal variation in abundance.—We measured changes in tiger abundance using tracking, camera trapping, and observational data over a 7-year period (1995–2002). We collected data year-round except July and August when heavy monsoon rains hindered fieldwork. For increased temporal definition, we recorded abundance in 2-month time segments following Ogutu and Dublin (2002). We assumed that in each time segment individual identity could be determined for all resident animals, and that all individuals present in the study area were detected.

Tigers can be classified as breeding or nonbreeding. Breeding tigers are resident male and female animals that have established territories to secure the resources necessary for successful reproduction (Smith and McDougal 1991; Smith et al. 1987, 1998b; Sunquist 1981). Adult tigers were considered resident if they were consistently detected in the same area at least 2 years in a row, or were observed with cubs.

Nonbreeding animals are cubs (0–1 years old), juveniles (1 year old until dispersal at 18–24 months), and transients. The distinction between cubs and juveniles is not particularly important in a biological sense, but is so in terms of monitoring because some camera-trap studies claim that detection of tigers < 1 year old is very low (Karanth and Nichols 1998; Karanth et al. 2004b; Kawanishi and Sunquist 2004; O'Brien et al. 2003). Juveniles that left their natal areas after 18 months of age were judged to have dispersed. Tigers that disappeared before that age were assumed to have died (Smith 1993). Transients were tigers that had dispersed from their natal areas but had not acquired a territory (preterritorial) or were forcefully ejected from their territories (postterritorial).

We assumed that transients in general do not breed and that females abandoning territories are not able to support dependent young. These assumptions are supported by previous observations on the behavior of wild tigers in Chitwan National Park (Smith and McDougal 1991; Smith et al. 1987; Sunquist 1981).

Tracking.—Tracking was carried out on a daily basis by 4 experienced field staff who covered all roads in the study area at least once per week by foot, bicycle, jeep, or elephant. They followed any fresh tiger tracks to find imprints clear enough for identification, from which they measured pad width, total width, and total length to ascertain age class and sex (McDougal 1999). Whenever possible, they identified individual tigers by track size in combination with the particular spatial arrangement of toes relative to the pad of each foot. Unambiguous identification required clear imprints of all 4 tiger tracks in good substrate (McDougal 1999). The number of cubs was deduced by the number of cub-sized track sets recorded with the tracks of a tigress. We estimated litter dates by cub tracks and known mating associations between resident animals. Some cubs may have died before we were able to detect their tracks and there could have been 1–2 months error in our estimates of some litter dates (Smith and McDougal 1991).

We conducted a blind field test of our assumption that we could identify individual tigers from their tracks (Smith et al. 1999). In this test we identified tigers (n = 17) passing through camera traps by their tracks and then checked against the photograph obtained; out of 54 sets of tracks analyzed, we identified 53 (98%) correctly. The only misidentified individual was a transient female mistaken for a juvenile. The same trackers were used for the duration of the study and continually (but informally) assessed each time a tiger walked through a camera trap. If unrecognized tracks were encountered, they were traced and the area was camera trapped. Ancillary data also were collected on marking sign because the concentration of scrapes and sprays can be used to infer territorial boundaries (Smith et al. 1998b).

Camera trapping.—We used Trailmaster camera traps (Goodson Associates, Lenexa, Kansas) consisting of a TM-1500 active infrared trail monitor linked with a TM-35 autofocus camera kit. Camera traps were set in pairs to take photos of both flanks of any passing tiger. Each pair was positioned on a likely tiger travel route, identified by the presence of tiger tracks, scrapes, and sprays (Karanth 1995). We covered the 100-km2 study area in 4 or 5 trapping blocks each year, with camera-trap locations chosen to maximize the number of tiger captures and to ensure that all tigers within the study area had a chance of being captured (Karanth and Nichols 1998). We trapped each block with groups of 4–10 camera traps spaced between 0.1 and 1.5 km apart. Each block was trapped 1–3 times per year and for 10–20 nights each time. We tested continuously our assumption that the study area was defined by the boundaries of the territories of resident female tigers, by camera trapping and tracking in adjoining areas.

Power analysis.—For monitoring, power is the probability of detecting a change should it occur (Cohen 1988; Gerrodette 1987). Power is affected by 5 parameters: sample size, variance, effect size, α (type I error), and β (type II error—Cohen 1988). Program TRENDS (Gerrodette 1993) facilitates calculating any 1 of these parameters given values for the other 4. We used TRENDS to calculate the overall minimum detectable negative change in tiger abundance (effect size) for monitoring programs conducted over 2–10 years. We set α and β at 0.2, and used the coefficient of variance (CV) in mean annual abundance (for all animals and resident animals alone) as the variance parameter. To best represent the natural changes in the tiger population, we assumed an exponential response and constant CV in relation to abundance (Hayward et al. 2002; Thompson et al. 1998). Within-sample variance, which would decrease power, was not estimated in the study and not included in the analysis.

Results

Effort and number of tigers detected.—Total effort in the 100-km2 study area over 7 years was 3,816 camera-trap nights (mean 545/year ± 251 SD, range 254–877/year) and approximately 1,900 days of tracking. Eleven resident females, 4 resident males, 20 transients, and 56 young (cubs and juveniles) were recorded (91 total; Fig. 1).

Fig. 1

Individual tigers (Panthera tigris) detected in the study area over 7 years. BF = breeding female, BM = breeding male, Y = young (cubs and juveniles), and T = transient. Sequential litters of offspring are listed directly under their respective mothers.

Some tigers matured over the course of the study so were counted in more than 1 demographic group. In total, we detected 82 individual tigers, of which we photographed 65 (79%; 631 photos), and detected 17 (21%) by tracking and observation. Every tiger counted as present in the study area was often detected multiple times within each 2-month period, and many of the resident tigers were recorded daily. All 17 animals not caught on film were cubs; 15 died before reaching 5 months of age, and 1 litter of 2 cubs dispersed from the study area before being photographed. The estimated mean age of young at 1st photographic capture was 13 months (range 5–20 months).

We obtained an additional 189 tiger photos from 1,665 camera-trap nights in the areas immediately adjacent to the study area. Nonetheless, females resident in the study area or their dependent cubs were never photographed or detected by their tracks in these peripheral zones. This result suggests an approximate mean density of 6 resident female tigers/100 km2 (ranges 4–7 resident female tigers/100 km2, CV = 0.10) and an approximate home-range size of 16.7 km2. We did not estimate home-range sizes for males because occasionally we detected them in peripheral zones. In general, both males and females showed strong site fidelity with little noticeable overlap between home ranges. High concentrations of spray and scrapes were observed along territorial boundaries, which tended to be delineated along roads, paths, or water courses.

However, 2 resident females moved territories; 1 left her territory to 1 of her female offspring and took over an adjoining area by displacing another female, and in another case a displaced female managed to reestablish herself in a neighboring territory made vacant by the death of the resident.

Trend in abundance.—Loss of resident animals, particularly males, caused large fluctuations in numbers of young. The decrease in overall tiger abundance from 23 in year 1 to 9 the next year was primarily due to the loss of 5 litters (total 12 cubs), all of which were <5 months old. Although not directly observed, circumstantial evidence (from sightings, tracks, and carcass examination) strongly suggested that all cubs were killed by new males acquiring territories from the previous residents. In year 2, part of the original male's territory was taken over by another male, but the following year both these males were usurped by 2 new males (Fig. 2). Only 1 litter of 2 cubs survived this period, which we attributed to the 1-year-old cubs’ relatively larger size at the time.

Fig. 2

Trend in tiger (Panthera tigris) abundance over 7 years. Abundance is estimated using tracking and camera trapping, assuming that all individuals in a 100-km2 study area are detected in each 2-month segment. No data were recorded in July-August because of the adverse weather conditions during the monsoon, a) Overall trends in total number of tigers (TOTAL), breeding tigers (B), and non-breeding tigers (NB). b) Trends in juveniles (J), cubs (C), and transients (T). c) Trends in breeding females (BF) and breeding males (BM).

After the initial social instability in years 2 and 3, the newly established males mated with 5 of the 6 resident females, who all produced litters in year 4. This synchronous litter production caused a sharp increase in tiger abundance to a peak of 24 in year 5. Tiger numbers then dropped as the young either dispersed outside the study area or died. Another peak of 19 tigers occurred in May–June of year 6, as the 5 resident females again produced litters within a short period (Fig. 2).

Apart from 8 individuals in year 6, transients were recorded at relatively low numbers. The peak occurred when several litters dispersed simultaneously from their natal areas but were still occasionally detected within the study area (Fig. 2).

Fates of tigers.—Of 11 resident females, 7 were still alive at the end of the study period, 2 disappeared after losing their territories to rivals, and 2 died (1 shot and 1 with an unexplained broken back). Of 4 resident males, 1 was still alive and 3 were displaced by rivals. Of the 3 displaced males, 1 disappeared from the study area, 1 was found dead after a fight with another tiger, and 1 was shot after becoming a man-eater.

Of 20 transient animals, 8 were preterritorial, 1 was postterritorial, 9 were of unknown origin, and 2 were residents whose territories lay immediately adjacent to the study area but were photographed (on 1 occasion each) by the perimeter camera traps (Table 1). One female transient was badly injured in a fight with another tiger, after which she returned to her natal area where she was found dead.

View this table:
Table 1

Demographic status of transient tigers (Panthera tigris) detected in the 100-km2 study area between 1995 and 2002. Roving residents were tigers who had territories adjacent to the study area but were detected, on 1 occasion each, by the perimeter camera traps.

PreterritorialPostterritorialRoving residentsUnknownTotal
Transient Females511411
Transient males30159

In addition to the 5 litters of cubs killed by infanticide, 2 litters (total 4 cubs) died because they were too young to fend for themselves when their mothers died. One juvenile tiger was presumed dead after being photographed with severe injuries from a deer snare. The remaining young lived long enough to reach dispersal age, 2 of them becoming residents in the study area.

Temporal variance in abundance.—Assuming that we detected all individuals present, mean abundance in the 100-km2 study area was 18 tigers (range 9–24, CV = 0.24). The high variation in abundance was due primarily to fluctuations in the number of nonbreeding tigers ( = 11, range = 2–16, CV = 0.39; Table 2). Conversely, the number of breeding animals remained relatively stable ( = 8, range = 6–9, CV = 0.09). At any time, generally 1 or 2 resident males and 6 resident females used the study area (Table 2). Overall, breeding tigers accounted for a mean 45% (range 30–78%, CV = 0.27) of total abundance, with resident males 10% (range 0–15%, CV = 0.37), and resident females 35% (range 23–67%, CV = 0.31). Nonbreeding tigers contributed a mean 55% (range 22–70%, CV = 0.21) to total abundance, with cubs 30% (range 0–65%, CV = 0.51), juveniles 18% (range 0–55%, CV = 0.83), and transients 7% (range 0–40%, CV = 1.17).

View this table:
Table 2

Variation in tiger (Panthera tigris) abundance between 2-month segments over 7 study years (1995–2002), based on tracking, camera trapping, and observational data. Abundance estimates are rounded up and derived from tracking and camera trapping for 10 months/year. We assume that all individual tigers present in each 2-month time period were detected.

CategoryN ̂RangeCV
Resident males20–20.33
Resident females64–70.10
Juveniles40–120.89
Cubs60–150.64
Transients20–81.25
Breeding animals86–90.09
Nonbreeding animals112–160.39
Tigers > 1 year old126–200.25
All tigers189–240.24

Power analysis.—Mean annual CV in abundance for breeding animals and all animals combined was 0.07 and 0.18, respectively. The minimum detectable negative change in abundance over 2–10 years calculated by TRENDS (1-tailed test, α = 0.2, β = 0.2), was 19–11% for breeding animals and 44-27% for all animals combined (Fig. 3).

Fig. 3

Minimum detectable overall change in tiger (Panthera tigris) abundance. Detectable change is calculated by TRENDS (Gerrodette 1993) using a 1-tailed test, α and β levels of 0.2, and the mean annual CV in estimated abundance for both breeding animals and all demographic groups combined.

Discussion

Tiger abundance.—The density estimate of 6 resident female tigers/100 km2 and the high overall abundance of 18 tigers recorded in this study suggest that the alluvial floodplain habitat of the Terai is one of the best tiger habitats anywhere in the world (Sanderson et al. 2006). This is supported by previous work in Chitwan (McDougal 1977; Seidensticker 1976; Smith 1993; Smith and McDougal 1991; Smith et al. 1987, 1998b; Sunquist 1981), and highlights the Terai Arc Landscape's potential as a long-term stronghold for tigers, and the need to reestablish breeding habitat and dispersal corridors across this region (Wikramanayake et al. 2004).

Changes in the number of nonbreeding animals caused most of the variation in local abundance, not changes in the number of resident adults, which remained relatively constant. These findings are similar to a study of lion (Panthern led) demography in Kenya, which reported stability in the number of breeding animals relative to nonbreeding individuals (Ogutu and Dublin 2002).

The large fluctuation in number of tiger young was due to infanticide that occurred when adult males acquired new territories. With male land tenure typically only about 2.8 years (Smith and McDougal 1991), infanticide and subsequent synchrony of litter births by resident females may be common and account for much of the observed oscillations in local abundance. Infanticide also has been observed for lions in the Serengeti, where average male tenure is 2.75 years and male takeover of prides results in significant increases in mortality of young followed by synchrony of litter births (Packer and Pusey 1983; Packer et al. 2001).

The method employed in our study to detect change in tiger abundance is difficult to replicate in other areas that do not have similarly suitable tracking media or highly skilled field staff. This study would have been improved by data collection strategies that allowed an estimate of the number of animals present but undetected. Considering our intensive monitoring effort, our confirmed ability to identify individuals from their tracks (Smith et al. 1999), and the stability of female home ranges (Smith et al. 1987), our assumption that we were able to detect all breeding tigers probably holds true. Some young cubs (0–2 months old) probably died before they could be detected, however, and numbers of transients also could have been underestimated (Smith and McDougal 1991).

Delineation of the study area was based on the assumption that it fully encompassed the collective home ranges of the resident females, and that there were no substantial changes in their territorial boundaries over the study period. Extensive camera trapping and tracking in adjoining areas did not record any of the study area females, suggesting that if changes in territorial boundaries did occur, they were relatively inconsequential.

Power to detect change.—Monitoring programs that identify the breeding sector of a population have considerably higher power to detect change than comparable efforts tracking total abundance. In addition, the minimum detectable change in abundance did not take into account within-sample variance, which will substantially decrease overall power.

However, the degree to which fluctuations in nonbreeding animals contribute to overall variance in tiger abundance is dependent on both spatial and temporal scale. Surveys that cover more territories of males will decrease the overall variance associated with normal turnover of male residents. Different time intervals between samples also will affect the variance; for abundance of breeding and nonbreeding tigers, mean CV was higher for 2-month versus 1-year intervals.

Implications.—Small-scale monitoring programs may have low power to detect trends, and if trends are detected they may not be useful to evaluate the effectiveness of conservation efforts. We recommend that tiger monitoring programs are designed to differentiate between the breeding and nonbreeding sections of the population. Identifying the residents will increase power to detect change and improve inferences regarding population status and long-term viability. In general, resident animals can be identified through camera trapping when recorded in the same area for 2 consecutive years, or if there is evidence of cubs accompanying females.

If residents cannot be identified, then monitoring should be carried out at a large enough scale that turnover of individual resident tigers will be unlikely to greatly affect detected changes in abundance. The size of survey area, in terms of the number of male territories and the length of time between samples, should be based on how much change needs to be detected for management purposes. Ideally in such cases, population modeling combined with data on sizes of tiger home ranges would be used to design the sampling approach. This study also supports previous work suggesting that small tiger populations are extremely vulnerable to extinction (Ahearn et al. 2001; Kenny et al. 1995; Smith and McDougal 1991). Loss of resident males increases cub mortality substantially through infanticide and temporarily limits the supply of individuals available for replacing residents who die. The loss of resident males in our study was due to displacement by competitors, but increased mortality of males due to poaching or disease also would suppress reproduction and make the population more susceptible to threats.

Acknowledgments

We thank the Department of National Parks and Wildlife Conservation for permission to carry out this study. T. Maskey, G. Upadhaya, and T. Adhikari provided valuable support and advice. The International Trust for Nature Conservation, Tiger Mountain, Fund for the Tiger, Himalayan Kingdoms, and Terre et Fauna made considerable financial and logistical contributions. We are greatly indebted to the camera trappers R. Mahato, I. Kumal, and Bir. B. Kumal for their tireless work under sometimes difficult and dangerous conditions. We were assisted in the field by many staff of Tiger Tops Jungle Lodge, including M. Cotton, J. Roberts, D. Bellas, and D. Thapa. We are grateful to U. Karanth, D. H. Johnson, T. Arnold, D. Siniff, F. Cuthbert, D. Garshelis, M. Perez, J. Slaght, P. Cutter, C. Greenwood, and 2 anonymous reviewers, who all made helpful comments on earlier drafts.

Footnotes

  • Associate Editor was Roger A. Powell.

Literature Cited

View Abstract