To decide whether an animal population is threatened or even on the brink of extinction, you need to know, at least approximately, how many individuals are left and if those numbers are increasing or decreasing. In other words, you can’t protect wildlife if you can’t monitor it.

Unfortunately, determining the number of animals that live in a certain habitat can be problematic. Current techniques, such as aerial counting, are expensive, time-consuming and prone to human error. Wildlife tallies can get distorted in other ways, too: by changes in climate or weather, by challenging terrain or by variations in an animal’s grouping behavior, range or timidity. In the case of rare species living in remote regions, scientists sometimes attempt to cobble together a picture of the population using multiple survey methods, such as automated wildlife cameras, DNA samples, tracks and other signs.

And, wild animals can sometimes be rather uncooperative. Take African elephants, for example. They inhabit changing and complicated landscapes—and they have a knack for covering themselves with mud.


It seems counterintuitive that lion populations have declined by 50 percent since 1994, the year that The Walt Disney Company released the very popular movie The Lion King. The film did focus conservation attention on African lions, but it wasn’t realistic regarding how a lion pride functions.

An additional pitfall is that we can even get in our own way, despite our best intentions. For instance, almost half of all wild lions have disappeared from the African continent since 1994, which is, coincidentally, the same year that The Walt Disney Company released the movie The Lion King. The film gave wild lions immense publicity and conservation attention, yet their populations continue to falter. It’s also not clear if people realize that the lion pride featured in The Lion King bears no resemblance to a real lion pride, which would eventually kick little Simba out—not crown him king.

Now, however, two new methods for counting animals are showing promise: machine learning and a novel technique called Spatially Explicit Capture-Recapture (SECR).

Counting elephants from a high elevation

In a recent paper that appeared in the science journal Remote Sensing in Ecology and Conservation, researchers report that they have harnessed machine learning to count African elephants from space with impressive accuracy.


Counting elephants is difficult. Their natural habitats are remote, varied and wild; and they often change the color of their skin by covering themselves with mud.

Using a computing system called Convolutional Neural Networks (CNN), they were able to automatically count elephants from satellite images, a method that is more reliable than aerial counts and much less labor-intensive.

Formerly, researchers used machine learning and satellite images to count a small handful of threatened species, such as albatrosses, seals and whales. But because the ocean is all blue, counting there is a lot less challenging. Tabulating elephants is trickier.

To program the CNN to detect elephants, the researchers first had to “train” the system using satellite images of elephants. They chose images taken in Addo Elephant National Park in South Africa between 2014 and 2019 by WorldView‐3 and WorldView‐4—satellites that capture some of the highest-resolution imagery commercially available. The researchers choose images from different seasons and years in a range of habitats, from dense shrublands and forests to grasslands and barrens.


We can now count elephants from space. Using satellite images is proving to be more reliable and much less labor-intensive than the former method of using aerial counts.

The scientists then tested how accurately the CNN program could detect elephants in the park against human volunteers screening the same images. The two detected elephants at similar rates: the CNN correctly identified elephants 75 percent of the time, whereas the human volunteers were successful 78 percent of the time. The true number of elephants was determined by two expert annotators. When the CNN’s detection was tested outside of the park, the accuracy dropped to 57 percent, which, the paper’s authors write, could be improved with a small amount of training data.

One big challenge for detecting animals using this method going forward is acquiring a large enough database to “train” a CNN, but crowdsourced labeling platforms, such as Zooniverse and Amazon Mechanical Turk, are promising tools. The authors conclude that “using high-resolution satellite imagery as a wildlife surveying tool will inevitably increase in the future as image resolution improves and costs fall.”

Locating lions from logging miles

According to World Wildlife Fund, there are about 23,000 lions left in the wild. Other calculations estimate that between 20,000 and 30,000 individuals are scattered among 102 populations spread across approximately 965,255 square miles of Africa. Historically, though, lions have disappeared from more than 90 percent of their range.


Using observed tracks to count wild lions in a particular area is simply not reliable enough to understand how the big cats are faring over time.

Most of these evaluations are based on audio-lure surveys (playbacks of recorded lion calls), expert solicitation and track counts, which are simply not reliable enough to understand how lion populations are doing over time. Now, an analytical method known as Spatially Explicit Capture-Recapture—which involves driving extensively and searching actively for lions and then taking high-quality photographs to individually identify them and noting their locations—is getting better results in counting big cats and understanding their movements.

The method was first tried in Uganda’s Queen Elizabeth Conservation Area, where lions spend a lot of their time up in trees and where it’s relatively easy to get good pictures of them. Due to this unique tree-climbing behavior, managers and tourists at the park frequently see lions. SECR experiments showed, however, that these lions are now moving more and have larger home-range sizes compared to the results from a previous study conducted about a decade ago. Since larger home-range sizes in big cats are usually associated with reduced density due to poorer prey availability, this is a concerning trend.

SECR researchers conclude that there’s great value in using methods that keep track of lion populations directly, and they urge conservation and research communities to cease using ad hoc, indirect techniques and shift to more direct and reliable strategies. Since the researchers published their study in 2020, SECR has been applied in the Maasai Mara and adopted by the Kenya Wildlife Service to survey lions and other carnivores across territories.

Lions in trees are easier to spot—and to photograph. ©Malcolm Cerfonteyn, flickr

Ticking off new terms of existence

The knowledge gained from tallying wildlife is vital for setting conservation goals and priorities. Counting and recording animals—which is the essence of wildlife monitoring—makes it possible to assess a population’s status and make decisions on whether action is needed to protect or recover the species.

Indeed, news about the loss of numbers of a particular species can raise concerns and often mobilize specific conservation actions. Such was the case when the Great Elephant Census spurred attendees at the 2016 International Union for Conservation of Nature World Conservation Congress to pass a motion urging all countries to close their domestic ivory markets.

So, if we can find new ways to total our wild neighbors so that we calculate how to safeguard them and live alongside them, I believe that’s a reckoning we desperately need to make.

Here’s to finding your true places and natural habitats,