Counting Prey Animals

Using drones, vertical camera traps, and AI spot pattern recognition to better estimate the prey base for tigers.
In Bardiya National Park (BNP), many species of large grazers, such as nilgai and sambar deer, have experienced a decline in numbers or even disappeared completely over the last couple of decades. As a result, the tiger population in BNP now relies mostly on chital deer (Axis axis), also known as the spotted deer.
See: Deer and Boar in Bardiya National Park.

The chital population seems to be under pressure due to changes in their habitat, as well as an increasing number of tigers that feed on them. This raises concerns over the status of the chital population and, consequently, the prey base for tigers. How many chital are there? Is the population decreasing?
In 2019 a network of vertical camera traps was installed in BNP, yielding millions of images.

The raw images are automatically classified to the species level using artificial intelligence (AI) with most animals identified as chital. 
The second step focuses on a distinctive feature of chital: their striking coat of spots. Each deer has a unique spot pattern, much like a human fingerprint. By capturing high-resolution images of these spot patterns, we can create a database of individual chital. However, with rough estimates of the chital population running into the thousands, a manual approach to such individual identification is not feasible. Here, AI can help us compare the spot patterns to build a large database with individually identified chital. Currently, the algorithm used to compare the spot pattern is being improved, for greater efficiency and precision. This increases the number of times an individual is sighted, reducing errors in the dataset and ultimately allowing for more comprehensive monitoring of the chital population. 

While the vertical camera traps have already provided extremely useful insights already, a recent data gap was identified. It seems that a proportion of the chital population mainly uses the grasslands of BNP and rarely ventures into the forest. Since the camera traps are only set up in the forest, no data is collected from these grasslands. In other words, are we capturing the entire chital population, or just a part of it? This is where drones come in. Unlike the camera traps, which continuously monitor a small patch, drones can capture large areas in a short amount of time. Therefore, they are a useful tool to complement the vertical camera traps by capturing entire herds of chital by flying over the grasslands periodically. The drones are equipped with high-resolution cameras and can capture images very similar to the camera traps, all without disturbing the chital. 

Finally, it is considered to equip about 15 chital with GPS collars. Following the activity and behavior of several chital will further complement the dataset. The chital are preferably selected from several herds that frequent different parts of BNP, ensuring sufficient coverage. Insights into where and when these chital are in the landscape will further our understanding of the entire population. With this data, we can make better estimates of the size of the chital population and start looking deeper into the chital’s behavior and answer questions such as: to what extent are chital affected by heat stress? Are they avoiding or preferring certain areas? How do the seasons affect their movement patterns? 

To read more about the research projects we have initiated in Bardiya National Park, click on:
Human-wildlife coexistence in Bardiya
Managing subtropical monsoon grasslands
Climate change
Re-wilding Bardiya National Park
Mountain Tiger (Snow Leopard)
“Save the Tiger! Save the Grasslands! Save the Water!”