SPOT THE SPOTS

The east African savannahs of the Serengeti and Tarangire ecosystems are among the most spectacular landscapes on the planet. These ecosystems in northern Tanzania are covered by grasslands and acacia trees, with millions of large mammals like zebras, lions, antelopes, elephants and leopards roaming freely. One of the most magnificent and iconic animals of these savannahs is the giraffe, the tallest mammal in the world.

Here, on an area of around 25,000 square kilometers, Monica Bond from the University of Zurich and her team at the Wild Nature Institute are conducting a comprehensive study of births, deaths, movements, and behaviors of Masai giraffes (G. c. tippelskirchi). The Masai Giraffe Project has been running since 2011 and is based on data from around 4,000 animals. It is the largest individual-based giraffe demographic study in the world.

CONSERVATION MEASURES FOR THE ANIMALS
Masai giraffes are threatened with extinction due to human activities. Over the last three decades, their population has declined by 50 percent. More and more of the savannah woodlands in which the animals live are being converted into agricultural land. Poaching is also a growing problem – Masai giraffes are killed for bushmeat markets.

The Masai Giraffe Project aims to help stop these negative developments. Monica Bond and her team record the births and deaths of their 4,000 study animals and observe how the giraffes interact with each other and how they move through their habitats. This is how they find out where the giraffes are doing well, where they are not, and why.

With their research, they provide important knowledge for the development of conservation measures that are necessary for the survival of the species. They have found, for example, that female giraffes that tend to group with more other giraffes survive better than socially isolated animals (→Bond et al., 2021), but that proximity to humans weakens their social cohesion (→Bond et al., 2020).

Monica Bond and her team are only able to obtain these data because they can distinguish individual animals from one another. Giraffes are ideal for individual-based studies like the Masai Giraffe Project – because just like our fingerprints, giraffe coat patterns are individually unique and unchanging throughout their lives.

A PATTERN RECOGNITION SOFTWARE
Since the 1960s, ecologists have used coat patterns to identify giraffe individuals. For a long time, researchers distinguished animals by eye based on their coat patterns, either from photographs or from coat pattern drawings they made in the field.

A little over ten years ago, the first pattern recognition software finally appeared which automated this work. Today, programs like WildID (→Bolger et al., 2012) make it possible to work with huge data sets. The combination of digital photography with machine learning opens entirely new research possibilities. For example, it can be used to analyze the social relationship networks of several thousand giraffes, which would be impossible without digital tools.

Monica Bond and her team work with WildID to distinguish the 4,000 giraffe individuals they study. Six times a year, they travel the 25,000 square kilometers of their study area along a set route, photographing every giraffe they see and noting its location, age, and sex. In just under two weeks per ecosystem, they usually collect more than a thousand photos, which they then feed into WildID for coat pattern recognition (→Lee, Bond et al., 2022).

To facilitate image recognition with WildID, the Wild Nature Institute, together with Microsoft AI for Good, have developed an automated process to crop all these photos to the giraffe torso. This saves researchers days or weeks of routine work, which they can instead spend analyzing their data (→Buehler, Lee et al., 2019).

A MEMORY FOR RESEARCH
Pattern recognition software programs are now the norm in giraffe research. For one thing, they are less expensive than other data collection methods: many of these programs are freely available online, and anyone with a camera and a computer can use them. Secondly, working with digital photos or camera trap recordings in combination with AIs is less invasive than other methods. Thanks to them, animals do not have to be captured and fitted with GPS trackers, for example, to collect data on individuals. Captures are time-consuming and extremely expensive, and also stressful for the animals – and it is important to Monica Bond and her team that they do not disturb the wild giraffes with their research.

With spot the spots, the University of Zurich has developed an interactive format that aims to communicate the possibilities which machine learning opens for wildlife research to a broad audience. Using images from the Masai Giraffe Project, scenographer Sonja Koch and graphic designer Meltem Kalayci designed a magnetic memory game that challenges children and adults to compete with pattern recognition technology.

This way, spot the spots conveys how artificial intelligence can be used in nature and species conservation – and why that’s important. After all, machine learning enables long-term and in-depth research projects in which researchers can answer more and different questions than they possibly could without digital tools.