Various guenons. Credit: William Allen, except 5th on bottom row (by M. Emetshu) and 6th on 3rd row (N. Rowe).

On the Origin of Colourful Monkey Faces

ByEd Yong
February 05, 2015
6 min read

“Can you hear that?” says James Higham.

I can. It’s a loud screeching noise in the background of our phone call.

“That’s a female rhesus freaking out,” he says.

Rhesus macaques have featured heavily in lab experiments, but this particular loud female is part of a wild group, living in Puerto Rico. Higham, an anthropologist from New York University, is studying them. He is interested in their faces, which vary from a dull pink to a vivid red. Specifically, he wants to know if the females judge the males on the intensity of that colour.

“I’m stood about 5 metres away from a sub-adult male and he’s with a 3-year-old female, and they’ve been mating a lot,” he says. “There are lots of other monkeys here, and the big, blue Caribbean sea around me.” As field work goes, it’s not hard.

The same couldn’t be said for the other group of funky-faced monkeys that Higham has been studying—the guenons. These African monkeys are known for their beautiful and diverse faces. De Brazza’s monkey has a white moustache and beard, and an orange sun rising on its forehead. The crowned guenon: dark eyeshadow, a black quiff, a pair of white forehead highlights, and a luxurious golden beard. The red-eared guenon: a drunk’s pink nose, a black brow ridge, white tufts around its eyes, and—yes—red ears. Every species of guenon, and there are between 24 and 36 of them, has its own distinctive facial marks.

Why?

LIMITED TIME OFFER

Get a FREE tote featuring 1 of 7 ICONIC PLACES OF THE WORLD

In the 1980, zoologist James Kingdon suggested that they recognise members of their own species by their faces. Many of these monkeys live in the same place, and some travel in large mixed groups. They live, feed, and watch out for predators together, but when it comes time to mate, their faces help them to find partners of their own kind.

The idea made sense; testing it has been difficult. For a start, guenons live in forest canopies and move quickly. It’s hard to look at their faces, let alone look at them looking at each other’s faces.

Their patterns are also complicated, so how do you objectively compare them? If two guenons have yellow sideburns and pink noses, but differently shaped brows and differently coloured eyes, are they similar? A bit different? Very different? Humans are terrible at this kind of task; we have to limit ourselves to comparing specific features, which Kingdon found frustrating.

Higham opted for a different approach. He and postdoc William Allen took hundreds of photos of 22 guenon species in various zoos and wildlife sanctuaries, and analysed them with the eigenface technique—a facial recognition programme developed in the 1980s. It can quantify how distinctive two faces are by comparing them across many features simultaneously.

The technique revealed that guenon species have more distinctive faces when they live together. This supports Kingdon’s hypothesis that the colourful facial palettes help neighbouring monkeys to recognise their own kind, despite sharing the same forests. By contrast, if the faces were adaptations to something in the monkeys’ environment—say, light levels—then species that live together should look more similar. In fact, it’s the opposite.

Next, Higham and Allen wanted to know if the monkeys could glean any more information from each other’s faces. Could guenons tell each other’s age or gender? Could they recognise individuals, as we humans can?

The duo used a computer programme to analyse 541 images from the same photo set, on the basis of either overall patterns or specific features—like the brightness, colour, shape, or size of their eyebrows and nose spots. They wanted to see if the programme could, based on these traits, classify the monkeys by species, age or gender, or recognise individuals. For example, after seeing photos of different monkeys, could the programme accurately identify one in a new photo? Likewise, after seeing photos of monkeys of both genders, could it tell if a new monkey was male or female?

The programme flunked the age and gender tests—there’s apparently nothing in a guenon’s face that reveals either characteristic. But it excelled at both species and individual recognition. The former isn’t surprising but the latter is.

As Kingdon suggested and Higham confirmed, guenons have evolved to look as different as possible when they live together. The need for differences between species ought to constrain the differences within them. “If you start looking very different from others of your species, you run the risk of being mistaken for something else,” says Higham. This kind of “stabilising selection” should lead to distinctive species but lookalike individuals—and yet that’s not what he found. One guenon can potentially tell its neighbours apart with a glance.

Of course, there’s no guarantee that what the programme is seeing is what the guenons actually see. Higham is now carrying out some experiments to see if the differences that the computer can glean are actually obvious to the monkeys themselves. He has even played around with drones to see if he can get a closer view of the guenons as they clamber through the canopy.

In the meantime, he thinks that his methods will be broadly useful to scientists who study animal signals. “There’s a lot of work on colour signals in animals and a lot of it is quite simple, like: This lizard patch varies from pale pink to bright pink,” he says. “But a lot of these signals are very complex. So, how do you measure them?” Computer learning offers an option. “That’s true whether it’s a guenon or a paper wasp or a paradise flycatcher or anything really.”

References: Allen, Stevens & Higham. 2014. Character displacement of Cercopithecini primate visual signals. Nature Communications. http://dx.doi.org/10.1038/ncomms5266

Allen & Higham. 2015. Assessing the potential information content of multicomponent visual signals: A machine learning approach. Proc Roy Soc B. http://dx.doi.org/10.1098/rspb.2014.2284

Go Further