Ecology and restoration of Sumatran tigers in forest and plantation landscapes

Tigers (Panthera tigris Linnaeus, 1758) are in danger of extinction. Their populations have declined from ~100,000 to only ~3,000 individuals in a century and their habitat has shrunk to less than 7% of the historic range. Of the five extant tiger subspecies, the Sumatran tiger (Panthera tigris sumatrae Pocock, 1929) is the most seriously threatened. Currently determined as Critically Endangered under IUCN criteria, the Sumatran tiger is likely to become extinct unless effective conservation measures are enacted. Threats to the tiger include habitat destruction, killing due to conflict with humans and livestock, and poaching for illegal wildlife trade.

Long-term survival of Sumatran tigers depends largely on the effectiveness of current conservation efforts in every tiger landscape. Successful conservation and management require accurate information on ecology of the species upon which decisions can be based. This study investigated basic ecological aspects of tigers and developed strategies for management and restoration to improve tiger viability in the Central Sumatra landscape. This landscape is comprised of natural forests and plantations managed for timber and agricultural commodities.

The first chapter assesses the variation in tiger abundance across forest types in Southern Riau, and over time in Tesso Nilo National Park, all in Central Sumatra. Using camera traps, my team and I systematically sampled five blocks representing three major forest types in the region: peat land, flat lowland, and hilly lowland. I found that tiger abundance varied by forest type and through time. Excluding two sampling blocks where no tigers were photographed, the lowest tiger density was in peat land forest of Kerumutan, and the highest density was in the flat lowland forest of Tesso Nilo. Repeated sampling in the newly established Tesso Nilo National Park documented a trend of increasing tiger density (SE) from 0.90 (0.38) individuals/100 km2 in 2005 to 1.70 (0.66) individuals/100 km2 in 2008. Overall, tiger densities from this study were lower than most previous estimates from other parts of Sumatra. The trend of increasing tiger density in Tesso Nilo, however, suggests that the tiger population could be augmented by protection of habitats that were previously logged and severely disturbed.

The second chapter examines the occupancy and habitat-use of the tiger across the major landcover types (natural forest, acacia plantation, oilpalm plantation, rubber plantation, and mixed agriculture). I found that tigers used some plantation areas, although they significantly preferred forests over plantations. In all landcover types, sites with tiger detections had thicker understory cover than sites without tiger detection. Modeling tiger occupancy while recognizing that probability of detection is not always perfect, I found that tiger occupancy covaried positively and significantly with altitude and negatively, but not significantly, with distance-to-forest-cores. Probability of habitat use by tigers covaried positively and significantly with understory cover and altitude, and negatively and significantly with human settlement and landcover rank. The results suggested that with adjustments in plantation management, tigers could use or roam through plantations within the habitat mosaic provided that the plantations had adequate understory cover and low level of human activity. They also could use riparian forests (as corridors) and smaller forest patches (as stepping stones) to travel between the main habitat patches across the forest and plantation landscape.

The third chapter investigates the ecological characteristics and possible inter-specific interactions among wild felids, including tigers and smaller cats, based on data collected using systematic camera trapping in combination with information on their natural history. I found that despite overlap in resource needs of the five felid species, each appears adapted to specific environmental conditions allowing coexistence with other felids. The five felid species used statistically different elevations, with the golden cat found to inhabit the highest elevation. Two-species occupancy models showed that only leopard cats were found to co-occur with other felid species more frequently than expected by chance under independence. Species of similar size or eating similar-sized prey generally tended to have low coefficients of temporal activity overlap, suggesting avoidance. Temporal avoidance is likely occurring in three pairs of felids, namely clouded leopards and golden cats, clouded leopards and marbled cats, and marbled cats and leopard cats. Based on the differences in morphological and ecological characteristics, and on patterns of spatial and temporal occurrence, I identified six possible mechanisms by which felids in Central Sumatra maintain coexistence. I discussed the implications of this study for management, focusing on how to balance diversity and abundance of felids.

The fourth chapter presents the tiger distribution models as a case study to illustrate the importance of accounting for uncertainty in species distribution mapping. I applied four modeling approaches, differing in how the response variable (tiger presence) is constructed and used in the models. I compared the performance and output of different models based on the relative importance of variables, descriptive statistics of the predictions, cross comparison between models using an error matrix, and validation using tiger presence data collected from independent surveys. All models consistently identified forest area within the grid as one of the most important variables explaining tiger probability of occurrence. Three models identified altitude as another important factor. While the four models were consistent in predicting relatively high probability of tiger occurrence for high elevation forest areas such as Rimbang Baling and Bukit Tigapuluh, they generally had a lower level of agreement in predictions for low elevation areas, particularly the peat land in the northeastern part of the study area. Based on the results of cross evaluation of the predictions among models and validation with the independent data, I considered the occupancy model to be superior to the others. If data collection format permits, I advocate the use of occupancy instead of the other modeling techniques to develop predictive species distribution maps.

The last chapter constructs a strategy to restore the tiger population across the ecosystem of Central Sumatra through integration of knowledge on tiger ecology from previous chapters with consideration of the ecological conditions of the landscape in the region. The strategy combines existing knowledge of tiger conservation and regional ecosystem restoration. It recognizes the limitations and challenges of traditional nature protection and considers existing and new opportunities. Emerging opportunities and new mechanisms, such as direct and indirect economic incentives for nature conservation and restoration, are taken into account. These, coupled with increased awareness of the stakeholders, better policies and implementation of good governance, and the willingness and know-how to maintain coexistence with wildlife among the local people, are expected to support and accelerate the recovery of tigers and their ecosystem.