Mimicry in action
By Michelle Catton

OTTAWA — Mimics come in many forms.

Tom Sherratt and his research student Arash Rashed study hoverflies to learn about natural mimicry.

Male fish pretend to be females to improve their chances of reproduction, non-poisonous beetles pretend to be poisonous by raising their backsides in the air and making idle threats to predators, and flies pretend they are wasps to avoid being eaten.

Batesian mimics, named after British naturalist Henry Walter Bates who first observed them in the mid-nineteenth century, are nature’s cheaters, piggybacks on the success of other species.

Their story begins innocently enough. It starts with two species, such as one type of butterfly and one type of moth, each hunted by the same predator.

The butterfly, like the Monarch, eats milkweed plants while the moth feeds on different plants. Since milkweed is poisonous and distasteful to birds, it makes the monarch unpalatable for its avian predators.

“Many organisms in nature are defended in some way. They are poisonous, or nasty tasting, or they have a bite or sting to deter predators. It’s better for these organisms if they can advertise themselves somehow, to say to the predators – hey, don’t eat me,” explains Christopher Beatty, a PhD student in behavioural ecology.

The Monarch butterfly next develops a brightly coloured pattern to warn predators of its poisonous nature, the beautiful orange and black design we see today. This is called the development of a warning signal, nature’s stop sign, telling predators to keep away.

The moth is now at a disadvantage. Not only is the moth a tasty treat for birds, its competitor for survival is advertising to predators its unpalatability, making the moth particularly vulnerable.

The birds have learned to avoid brightly coloured butterflies, so through natural selection, moths with bright colours resembling the Monarch live longer and produce more offspring than those with no pattern or colour.

Eventually, the moth develops a black and orange design strikingly similar to the Monarch butterfly. Even though it isn’t defended in any way, it uses the Monarch’s pattern as a shield, tricking predators into thinking it’s distasteful.

Arash Rashed, one of Sherratt’s PhD students, sits at the computer closest to the window. He is waiting, perhaps more eagerly than the rest of us, for spring to arrive in Ottawa. With the onset of warm weather, he can continue the field research he started last year, the study of the evolution of Batesian mimics, or copycats.

“I always have to wait for summer to do my experiments,” laments Rashed, who is examining the effects of insects on the evolution of mimicry, specifically dragonflies.

Although his research interests sound impressive, his experiments are less than glamourous. For days on end, Rashed stands in a field holding a stick above his head, dangling flies and bees in an attempt to entice predators.

“People might not think you could go fishing for dragonflies,” laughs Rashed, who developed strong arm muscles after “fishing” for six hours a day all summer long.

Rashed fished for dragonflies using hoverflies and bees as bait last summer.

Rashed is studying an area of evolutionary theory that has rarely been examined. Scientists have identified birds as the predators that influence the evolution of mimicry, but Sherratt and Rashed think insect predators, like dragonflies, can also play a role. They are examining whether or not they avoid eating brightly coloured, patterned species.

“Invertebrate (insect) predators can play a role in evolution as well. We wanted to know what role dragonflies play,” says Rashed.

They started with bees and hoverflies. Hoverflies are closely related to black flies, but they are bee mimics. They have the same body shape and black and yellow pattern as bees or wasps. Because hoverflies are such good mimics, humans can barely tell the difference between the two. Rashed wanted to know if dragonflies could sense something humans can’t.

First, he captured bees and hoverflies with insect nets, then hung one of each on strings descending from his stick. He recorded the results of his “fishing” — how often a dragonfly ate the bee offering compared to the number of times it chose the hoverfly.

“We saw a small tendency for the dragonflies to go for the mimics (hoverflies), but since it was very slight, we can’t be sure,” says Rashed. “Other studies have found that dragonflies avoid black and yellow in different species altogether.”

Because their findings are contrary to those published in the past, Sherratt and Rashed have decided to repeat their experiment this summer. They want to make sure their results are correct before they release them to the scientific world.

Although Rashed is eagerly awaiting spring and the chance to resume his field work, he keeps considerably busy over the winter months. In fact, he and Sherratt have pioneered a computer program that may change the way mimics are investigated forever. They call their program Neural Networks.

“Now, we have no way to objectively measure the similarity between mimics and their models. Sometimes, mimicry is not perfect, and we want to take that into account,” says Rashed.

A problem with mimicry field work done around the world is evaluating mimics. When Rashed chose hoverflies for his experiment, they all seemed like good mimics to him, but in fact, some hoverflies resemble bees more than others. Although the human eye may not be able to tell the difference, dragonflies probably can. Up until now, there was no way for humans to differentiate between a poor mimic and a near-perfect mimic. The Neural Networks computer program could change that.

“The system measures the antenna length, colour, length of wings, etc. of photographs of mimics and their models. We train the system to tell us how closely the mimic resembles the model,” says Rashed, whose program spits out a number to tell the researcher how good the mimic actually is. A near-perfect hoverfly can be close to 99 per cent similar to a bee, while a poor mimic might only be 50 per cent similar.

Rashed works on a mimicry computer program during the winter months.

Although the program is still in its beginning stages, Rashed and Sherratt are bolstered by comparisons with animal experiments. A study done at the University of Nottingham, in the United Kingdom, trained pigeons to differentiate between poor mimics and near-perfect mimics.

Their method was similar to that employed by Rashed’s computer program. They trained one set of pigeons to identify pictures of flies by rewarding correct identifications. They trained another set to identify bees. Then they presented all the pigeons with pictures of hoverflies, and recorded how often they were correctly identified as flies by the fly-trained pigeons, and how often they were mistakenly identified as bees by the bee-trained pigeons. They recorded percentages similar to those produced by Rashed’s program.

“The pigeons identified what looked like a poor mimic to humans as a near-perfect mimic. The computer program did the same thing,” says Rashed. This suggests his program can think like a pigeon, a predator, when humans can’t. The technology could be used by scientists all over the world investigating the evolutionary theory behind mimicry.

Christopher Beatty, who occupies a computer in the dusty, windowless corner of Sherratt’s lab, is working at the other end of the mimicry spectrum. While Rashed is examining detailed ways of evaluating mimics, Beatty wants to understand the basic systems behind the phenomenon. He uses humans as predators, monitoring their behaviour to simulate natural conditions. He looks for generalizations about mimicry.

“We’re not trying to evaluate human psychology,” says Beatty.

The phenomenon Beatty is evaluating is called Müllerian mimicry, named after a German zoologist. It happens when two or more species, both poisonous or defended, develop the same brightly-coloured pattern to warn predators away.

Beatty and Sherratt discuss Mullerian mimicry in their lab over morning coffee.

"It happens in the Amazon with poisonous frogs. Different frog species with the same predator are all the same colour. It makes sense for survival,” says Beatty.

For warning signals to work, predators must learn that a particular colour or pattern should be avoided. They have to eat one or two frogs before they learn. If all the frogs had different patterns, a predator would have to eat several of each variety before they learned them all. By maintaining a similar warning signal across species, like bright blue skin, a predator only has to eat one blue-skinned frog of one species before it learns that all blue frogs are bad.

Rashed, Beatty, and Sherratt examine data to uncover the mystery of mimicry evolution.

While Rashed fishes for dragonflies to test his theories, Beatty fishes for students to test Müllerian theory. He hovers in the library with his laptop, asking students to play his evolutionary computer game. The students hunt for computer-generated prey using a mouse, getting points for “eating” non-poisonous victims, and losing points when eating poisonous ones. Prey vary in their conspicuousness, and their physical traits are passed on to their digital offspring. In this way, Beatty can evaluate how warning signals evolve over time, and how they are passed on from generation to generation.

“We want to know how they get over the hump genetically, how they all develop the same warning signal,” says Beatty.

Most of the theories Beatty tests have been long-accepted in the scientific community, but because evolution is naturally slow, it is hard to observe it in nature. Beatty’s program allows these theories to be tested quickly and with relative accuracy.

“We know it’s artificial to use humans as predators, but we are examining relatively simple systems,” qualifies Beatty. “To a certain extent, it makes sense.”

After examining their data, Sherratt and Beatty have found that defended prey, those with warning signals, move slower and more erratically than undefended prey, giving predators more time to recognize their stop signs. This explains the dizzy flight pattern of the Monarch butterfly. They also found that defended prey evolve warning signals for the sole purpose of making undefended prey vulnerable.

“We find our results funny,” says Beatty. “We think – isn’t this the most logical way for this to work? But we have proven that this is the way it definitely works.”

Sherratt has a reputation as a theorist. His contribution to mimicry theory over the past decade has been considerable.

“He's an unusual combination of an excellent fieldworker and a researcher with significant mathematical and modelling skills - just the job in today's cut-throat academia,” says Francis Gilbert, a researcher in the U.K. also studying hoverflies. “In my view, his work has made most of the really important recent advances in the field, a field that has been more or less stagnant for decades.”

Although his work has few short-term practical applications, Sherratt says it’s important to know as much as possible about the natural world.

“It’s helpful to understand what makes evolution tick,” says Sherratt.