Bats are fascinating creatures that have been around for millions of years. They are the only mammals that can fly, and they have adapted to a wide range of habitats and lifestyles.
Bats can either be insectivorous, frugivorous, nectarivorous, or even carnivorous.
Some bats live in caves, some in trees, some in buildings, and some in other places. There are bats that are solitary, but some are social, and other bat species even form huge colonies that can number in the millions.
One of the most remarkable bat colonies in the world is found in Zambia, a landlocked country in southern Africa.
Every year, millions of African straw-colored fruit bats (Eidolon helvum) migrate to Kasanka National Park, where they roost in a small patch of forest.
They create one of the most spectacular wildlife displays on Earth, as they fly out every evening to feed on fruits and flowers.
They also provide important ecological and economic benefits to the region, as they disperse seeds and pollen over vast distances.
The wildlife wonder of Kasanka National Park
NARINDER NANU/AFP via Getty Images
Every year, from October to January, a spectacular phenomenon occurs in Kasanka National Park in Zambia: the arrival and departure of millions of African straw-colored fruit bats.
These bats are the only long-distance migrant fruit bats on the continent, and they form one of the largest and most important bat colonies in the world.
They roost in a small patch of forest in the park, and every evening they emerge from their roost site in a breathtaking display of flight and sound.
The bats fly out to feed on fruits and flowers, dispersing seeds and pollen over vast distances. They provide unparalleled ecosystem services that benefit both nature and people.
But how many bats are there exactly? And why is it important to know? These are the questions that have puzzled researchers and visitors for decades.
Past estimates based on visual counts have ranged from 1 million to 10 million, a sign of how difficult the task is.
To crack the problem, researchers needed a new approach: using artificial intelligence to count the bats.
Moreover, the counting has to be done in a short time window, before it gets too dark to see. The counting also has to be done in a way that does not disturb the bats or their habitat.
To overcome these challenges, researchers from Germany, Zambia, and the United States decided to use an array of small video cameras to film the bat emergence in a standard way, count the bats in each video, and then extrapolate a total number.
They used four cameras mounted on tripods at different angles and distances from the roost site. They filmed the bats for about 15 minutes every evening for five days in November 2019.
The next step was to count the bats in each video. This was also not an easy task, as the videos were noisy, blurry, and full of moving objects.
To automate the counting process, the researchers developed an artificial intelligence algorithm that could detect and track individual bats in each video frame.
The algorithm used a deep neural network, a type of machine learning technique that can learn from data and perform complex tasks.
The researchers trained the neural network on a subset of video frames that they manually annotated with bat locations. It then learned to recognize bats in new video frames and count them accurately.
The final step was to combine the counts from each video and extrapolate a total number for the whole colony.
The researchers used statistical methods to account for variations in bat density, flight direction, camera position, and other factors. They also compared their results with previous estimates based on visual counts and acoustic methods to monitor the growth or decline of the bat colony.
